sample research journal article

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings
• Advanced Search
• Journal List
• Ind Psychiatry J
• v.19(1); Jan-Jun 2010

Statistics without tears: Populations and samples

Amitav banerjee.

Department of Community Medicine, D Y Patil Medical College, Pune, India

Suprakash Chaudhury

1 Department of Psychiatry, RINPAS, Kanke, Ranchi, India

Research studies are usually carried out on sample of subjects rather than whole populations. The most challenging aspect of fieldwork is drawing a random sample from the target population to which the results of the study would be generalized. In actual practice, the task is so difficult that some sampling bias occurs in almost all studies to a lesser or greater degree. In order to assess the degree of this bias, the informed reader of medical literature should have some understanding of the population from which the sample was drawn. The ultimate decision on whether the results of a particular study can be generalized to a larger population depends on this understanding. The subsequent deliberations dwell on sampling strategies for different types of research and also a brief description of different sampling methods.

Research workers in the early 19th century endeavored to survey entire populations. This feat was tedious, and the research work suffered accordingly. Current researchers work only with a small portion of the whole population (a sample) from which they draw inferences about the population from which the sample was drawn.

This inferential leap or generalization from samples to population, a feature of inductive or empirical research, can be full of pitfalls. In clinical medicine, it is not sufficient merely to describe a patient without assessing the underlying condition by a detailed history and clinical examination. The signs and symptoms are then interpreted against the total background of the patient's history and clinical examination including mental state examination. Similarly, in inferential statistics, it is not enough to just describe the results in the sample. One has to critically appraise the real worth or representativeness of that particular sample. The following discussion endeavors to explain the inputs required for making a correct inference from a sample to the target population.

TARGET POPULATION

Any inferences from a sample refer only to the defined population from which the sample has been properly selected. We may call this the target population. For example, if in a sample of lawyers from Delhi High Court it is found that 5% are having alcohol dependence syndrome, can we say that 5% of all lawyers all over the world are alcoholics? Obviously not, as the lawyers of Delhi High Court may be an institution by themselves and may not represent the global lawyers′ community. The findings of this study, therefore, apply only to Delhi High Court lawyers from which a representative sample was taken. Of course, this finding may nevertheless be interesting, but only as a pointer to further research. The data on lawyers in a particular city tell us nothing about lawyers in other cities or countries.

POPULATIONS IN INFERENTIAL STATISTICS

In statistics, a population is an entire group about which some information is required to be ascertained. A statistical population need not consist only of people. We can have population of heights, weights, BMIs, hemoglobin levels, events, outcomes, so long as the population is well defined with explicit inclusion and exclusion criteria. In selecting a population for study, the research question or purpose of the study will suggest a suitable definition of the population to be studied, in terms of location and restriction to a particular age group, sex or occupation. The population must be fully defined so that those to be included and excluded are clearly spelt out (inclusion and exclusion criteria). For example, if we say that our study populations are all lawyers in Delhi, we should state whether those lawyers are included who have retired, are working part-time, or non-practicing, or those who have left the city but still registered at Delhi.

Use of the word population in epidemiological research does not correspond always with its demographic meaning of an entire group of people living within certain geographic or political boundaries. A population for a research study may comprise groups of people defined in many different ways, for example, coal mine workers in Dhanbad, children exposed to German measles during intrauterine life, or pilgrims traveling to Kumbh Mela at Allahabad.

GENERALIZATION (INFERENCES) FROM A POPULATION

When generalizing from observations made on a sample to a larger population, certain issues will dictate judgment. For example, generalizing from observations made on the mental health status of a sample of lawyers in Delhi to the mental health status of all lawyers in Delhi is a formalized procedure, in so far as the errors (sampling or random) which this may hazard can, to some extent, be calculated in advance. However, if we attempt to generalize further, for instance, about the mental statuses of all lawyers in the country as a whole, we hazard further pitfalls which cannot be specified in advance. We do not know to what extent the study sample and population of Delhi is typical of the larger population – that of the whole country – to which it belongs.

The dilemmas in defining populations differ for descriptive and analytic studies.

POPULATION IN DESCRIPTIVE STUDIES

In descriptive studies, it is customary to define a study population and then make observations on a sample taken from it. Study populations may be defined by geographic location, age, sex, with additional definitions of attributes and variables such as occupation, religion and ethnic group.[ 1 ]

Geographic location

In field studies, it may be desirable to use a population defined by an administrative boundary such as a district or a state. This may facilitate the co-operation of the local administrative authorities and the study participants. Moreover, basic demographic data on the population such as population size, age, gender distribution (needed for calculating age- and sex-specific rates) available from census data or voters’ list are easier to obtain from administrative headquarters. However, administrative boundaries do not always consist of homogenous group of people. Since it is desirable that a modest descriptive study does not cover a number of different groups of people, with widely differing ways of life or customs, it may be necessary to restrict the study to a particular ethnic group, and thus ensure better genetic or cultural homogeneity. Alternatively, a population may be defined in relation to a prominent geographic feature, such as a river, or mountain, which imposes a certain uniformity of ways of life, attitudes, and behavior upon the people who live in the vicinity.

If cases of a disease are being ascertained through their attendance at a hospital outpatient department (OPD), rather than by field surveys in the community, it will be necessary to define the population according to the so-called catchment area of the hospital OPD. For administrative purposes, a dispensary, health center or hospital is usually considered to serve a population within a defined geographic area. But these catchment areas may only represent in a crude manner with the actual use of medical facilities by the local people. For example, in OPD study of psychiatric illnesses in a particular hospital with a defined catchment area, many people with psychiatric illnesses may not visit the particular OPD and may seek treatment from traditional healers or religious leaders.

Catchment areas depend on the demography of the area and the accessibility of the health center or hospital. Accessibility has three dimensions – physical, economic and social.[ 2 ] Physical accessibility is the time required to travel to the health center or medical facility. It depends on the topography of the area (e.g. hill and tribal areas with poor roads have problems of physical accessibility). Economic accessibility is the paying capacity of the people for services. Poverty may limit health seeking behavior if the person cannot afford the bus fare to the health center even if the health services may be free of charge. It may also involve absence from work which, for daily wage earners, is a major economic disincentive. Social factors such as caste, culture, language, etc. may adversely affect accessibility to health facility if the treating physician is not conversant with the local language and customs. In such situations, the patient may feel more comfortable with traditional healers.

Ascertainment of a particular disease within a particular area may be incomplete either because some patient may seek treatment elsewhere or some patients do not seek treatment at all. Focus group discussions (qualitative study) with local people, especially those residing away from the health center, may give an indication whether serious underreporting is occurring.

When it is impossible to relate cases of a disease to a population, perhaps because the cases were ascertained through a hospital with an undefined catchment area, proportional morbidity rates may be used. These rates have been widely used in cancer epidemiology where the number of cases of one form of cancer is expressed as a proportion of the number of cases of all forms of cancer among patients attending the same hospital during the same period.

POPULATIONS IN ANALYTIC STUDIES

Case control studies.

As opposed to descriptive studies where a study population is defined and then observations are made on a representative sample from it, in case control studies observations are made on a group of patients. This is known as the study group , which usually is not selected by sampling of a defined larger group. For instance, a study on patients of bipolar disorder may include every patient with this disorder attending the psychiatry OPD during the study period. One should not forget, however, that in this situation also, there is a hypothetical population consisting of all patients with bipolar disorder in the universe (which may be a certain region, a country or globally depending on the extent of the generalization intended from the findings of the study). Case control studies are often carried out in hospital settings because this is more convenient and accessible group than cases in the community at large. However, the two groups of cases may differ in many respects. At the outset of the study, it should be deliberated whether these differences would affect the external validity (generalization) of the study. Usually, analytic studies are not carried out in groups containing atypical cases of the disorder, unless there is a special indication to do so.

Populations in cohort studies

Basically, cohort studies compare two groups of people (cohorts) and demonstrate whether or not there are more cases of the disease among the cohort exposed to the suspected cause than among the cohort not exposed. To determine whether an association exists between positive family history of schizophrenia and subsequent schizophrenia in persons having such a history, two cohorts would be required: first, the exposed group, that is, people with a family history of mental disorders (the suspected cause) and second, the unexposed group, that is, people without a family history of mental disorders. These two cohorts would need to be followed up for a number of years and cases of schizophrenia in either group would be recorded. If a positive family history is associated with development of schizophrenia, then more cases would occur in the first group than in the second group.

The crucial challenges in a cohort study are that it should include participants exposed to a particular cause being investigated and that it should consist of persons who can be followed up for the period of time between exposure (cause) and development of the disorder. It is vital that the follow-up of a cohort should be complete as far as possible. If more than a small proportion of persons in the cohort cannot be traced (loss to follow-up or attrition), the findings will be biased , in case these persons differ significantly from those remaining in the study.

Depending on the type of exposure being studied, there may or may not be a range of choice of cohort populations exposed to it who may form a larger population from which one has to select a study sample. For instance, if one is exploring association between occupational hazard such as job stress in health care workers in intensive care units (ICUs) and subsequent development of drug addiction, one has to, by the very nature of the research question, select health care workers working in ICUs. On the other hand, cause effect study for association between head injury and epilepsy offers a much wider range of possible cohorts.

Difficulties in making repeated observations on cohorts depend on the length of time of the study. In correlating maternal factors (pregnancy cohort) with birth weight, the period of observation is limited to 9 months. However, if in a study it is tried to find the association between maternal nutrition during pregnancy and subsequent school performance of the child, the study will extend to years. For such long duration investigations, it is wise to select study cohorts that are firstly, not likely to migrate, cooperative and likely to be so throughout the duration of the study, and most importantly, easily accessible to the investigator so that the expense and efforts are kept within reasonable limits. Occupational groups such as the armed forces, railways, police, and industrial workers are ideal for cohort studies. Future developments facilitating record linkage such as the Unique Identification Number Scheme may give a boost to cohort studies in the wider community.

A sample is any part of the fully defined population. A syringe full of blood drawn from the vein of a patient is a sample of all the blood in the patient's circulation at the moment. Similarly, 100 patients of schizophrenia in a clinical study is a sample of the population of schizophrenics, provided the sample is properly chosen and the inclusion and exclusion criteria are well defined.

To make accurate inferences, the sample has to be representative. A representative sample is one in which each and every member of the population has an equal and mutually exclusive chance of being selected.

Sample size

Inputs required for sample size calculation have been dealt from a clinical researcher's perspective avoiding the use of intimidating formulae and statistical jargon in an earlier issue of the journal.[ 1 ]

Target population, study population and study sample

A population is a complete set of people with a specialized set of characteristics, and a sample is a subset of the population. The usual criteria we use in defining population are geographic, for example, “the population of Uttar Pradesh”. In medical research, the criteria for population may be clinical, demographic and time related.

• Clinical and demographic characteristics define the target population, the large set of people in the world to which the results of the study will be generalized (e.g. all schizophrenics).
• The study population is the subset of the target population available for study (e.g. schizophrenics in the researcher's town).
• The study sample is the sample chosen from the study population.

METHODS OF SAMPLING

Purposive (non-random samples).

• Volunteers who agree to participate
• Snowball sample, where one case identifies others of his kind (e.g. intravenous drug users)
• Convenient sample such as captive medical students or other readily available groups
• Quota sampling, at will selection of a fixed number from each group
• Referred cases who may be under pressure to participate
• Haphazard with combination of the above methods

Non-random samples have certain limitations. The larger group (target population) is difficult to identify. This may not be a limitation when generalization of results is not intended. The results would be valid for the sample itself (internal validity). They can, nevertheless, provide important clues for further studies based on random samples. Another limitation of non-random samples is that statistical inferences such as confidence intervals and tests of significance cannot be estimated from non-random samples. However, in some situations, the investigator has to make crucial judgments. One should remember that random samples are the means but representativeness is the goal. When non-random samples are representative (compare the socio-demographic characteristics of the sample subjects with the target population), generalization may be possible.

Random sampling methods

Simple random sampling.

A sample may be defined as random if every individual in the population being sampled has an equal likelihood of being included. Random sampling is the basis of all good sampling techniques and disallows any method of selection based on volunteering or the choice of groups of people known to be cooperative.[ 3 ]

In order to select a simple random sample from a population, it is first necessary to identify all individuals from whom the selection will be made. This is the sampling frame. In developing countries, listings of all persons living in an area are not usually available. Census may not catch nomadic population groups. Voters’ and taxpayers’ lists may be incomplete. Whether or not such deficiencies are major barriers in random sampling depends on the particular research question being investigated. To undertake a separate exercise of listing the population for the study may be time consuming and tedious. Two-stage sampling may make the task feasible.

The usual method of selecting a simple random sample from a listing of individuals is to assign a number to each individual and then select certain numbers by reference to random number tables which are published in standard statistical textbooks. Random number can also be generated by statistical software such as EPI INFO developed by WHO and CDC Atlanta.

Systematic sampling

A simple method of random sampling is to select a systematic sample in which every n th person is selected from a list or from other ordering. A systematic sample can be drawn from a queue of people or from patients ordered according to the time of their attendance at a clinic. Thus, a sample can be drawn without an initial listing of all the subjects. Because of this feasibility, a systematic sample may have some advantage over a simple random sample.

To fulfill the statistical criteria for a random sample, a systematic sample should be drawn from subjects who are randomly ordered. The starting point for selection should be randomly chosen. If every fifth person from a register is being chosen, then a random procedure must be used to determine whether the first, second, third, fourth, or fifth person should be chosen as the first member of the sample.

Multistage sampling

Sometimes, a strictly random sample may be difficult to obtain and it may be more feasible to draw the required number of subjects in a series of stages. For example, suppose we wish to estimate the number of CATSCAN examinations made of all patients entering a hospital in a given month in the state of Maharashtra. It would be quite tedious to devise a scheme which would allow the total population of patients to be directly sampled. However, it would be easier to list the districts of the state of Maharashtra and randomly draw a sample of these districts. Within this sample of districts, all the hospitals would then be listed by name, and a random sample of these can be drawn. Within each of these hospitals, a sample of the patients entering in the given month could be chosen randomly for observation and recording. Thus, by stages, we draw the required sample. If indicated, we can introduce some element of stratification at some stage (urban/rural, gender, age).

It should be cautioned that multistage sampling should only be resorted to when difficulties in simple random sampling are insurmountable. Those who take a simple random sample of 12 hospitals, and within each of these hospitals select a random sample of 10 patients, may believe they have selected 120 patients randomly from all the 12 hospitals. In statistical sense, they have in fact selected a sample of 12 rather than 120.[ 4 ]

Stratified sampling

If a condition is unevenly distributed in a population with respect to age, gender, or some other variable, it may be prudent to choose a stratified random sampling method. For example, to obtain a stratified random sample according to age, the study population can be divided into age groups such as 0–5, 6–10, 11–14, 15–20, 21–25, and so on, depending on the requirement. A different proportion of each group can then be selected as a subsample either by simple random sampling or systematic sampling. If the condition decreases with advancing age, then to include adequate number in the older age groups, one may select more numbers in older subsamples.

Cluster sampling

In many surveys, studies may be carried out on large populations which may be geographically quite dispersed. To obtain the required number of subjects for the study by a simple random sample method will require large costs and will be cumbersome. In such cases, clusters may be identified (e.g. households) and random samples of clusters will be included in the study; then, every member of the cluster will also be part of the study. This introduces two types of variations in the data – between clusters and within clusters – and this will have to be taken into account when analyzing data.

Cluster sampling may produce misleading results when the disease under study itself is distributed in a clustered fashion in an area. For example, suppose we are studying malaria in a population. Malaria incidence may be clustered in villages having stagnant water collections which may serve as a source of mosquito breeding. In villages without such water stagnation, there will be lesser malaria cases. The choice of few villages in cluster sampling may give erroneous results. The selection of villages as a cluster may be quite unrepresentative of the whole population by chance.[ 5 ]

Lot quality assurance sampling

Lot quality assurance sampling (LQAS), which originated in the manufacturing industry for quality control purposes, was used in the nineties to assess immunization coverage, estimate disease prevalence, and evaluate control measures and service coverage in different health programs.[ 6 ] Using only a small sample size, LQAS can effectively differentiate between areas that have or have not met the performance targets. Thus, this method is used not only to estimate the coverage of quality care but also to identify the exact subdivisions where it is deficient so that appropriate remedial measures can be implemented.

The choice of sampling methods is usually dictated by feasibility in terms of time and resources. Field research is quite messy and difficult like actual battle. It may be sometimes difficult to get a sample which is truly random. Most samples therefore tend to get biased. To estimate the magnitude of this bias, the researcher should have some idea about the population from which the sample is drawn. In conclusion, the following quote cited by Bradford Hill[ 4 ] elegantly sums up the benefit of random sampling:

…The actual practice of medicine is virtually confined to those members of the population who either are ill, or think they are ill, or are thought by somebody to be ill, and these so amply fill up the working day that in the course of time one comes unconsciously to believe that they are typical of the whole. This is not the case. The use of a random sample brings to light the individuals who are ill and know they are ill but have no intention of doing anything about it, as well as those who have never been ill, and probably never will be until their final illness. These would have been inaccessible to any other method of approach but that of the random sample… . J. H. Sheldon

Source of Support: Nil.

Conflict of Interest: None declared.

Journal Article: Introduction

When to write the introduction.

• Introduction

Your paper’s introduction is an opportunity to provide readers with the background necessary to understand your paper : the status of knowledge in your field, the question motivating your work and its significance, how you sought to answer that question (methods), and your main findings. A well-written introduction will broaden your readership by making your findings accessible to a larger audience.

Introduction Formula

Clarity is achieved by providing information in a predictable order.  Successful introductions are therefore composed of 4 ordered components which are referred to as the “introduction formula”.

• General Background. Introduce the general area of science in which your project takes place, highlighting the status of our understanding of that system.
• Specific Background. Narrow down to the sub-area that your paper will be addressing, and again highlight the extent of our understanding in this sub-area.

Tip: Give your readers the technical details they need to understand the system –nothing more. Your purpose is not to showcase the breadth of your knowledge but instead to give readers all the tools they need to understand your results and their significance.

• Knowledge Gap. After discussing what we know, articulate what we do not know, specifically focusing on the question that has motivated your work. The prior two components should serve as a set-up for this question. That is, the question motivating your work should be a logical next step given what you’ve described in the general and specific background.
• “Here we show…” Very briefly summarize your methods and findings. Note that you may end this section with a sentence or two on the implications/novelty of your results, although this is not essential given that you will more thoroughly address these points in the discussion section.

This content was adapted from from an article originally created by the  MIT Biological Engineering Communication Lab .

Resources and Annotated Examples

Annotated example 1.

Introduction from an article published in Science Translational Medicine . 4 MB

Annotated Example 2

Introduction from an article published in Cell . 2 MB

The Journal of Integrative Behavioral Science Writing Guide: Undergraduate Research Journal Examples

• Journal Requirements
• Journal Process
• Course Outline and Requirements
• Institutional Review Board
• Scientific Method
• Article Example
• Journal Articles
• Types of Research Articles
• About Zotero
• APA Citations
• Books & e-books
• Paraphrasing
• Web Resources
• Research Management Systems
• Undergraduate Research Journal Examples

Psychology Journals

• Journal of Psychological Inquiry "The journal exists for a variety of reasons, and a primary one is to illustrate the high quality of undergraduates' scholarly work. Scholarly work encompasses a broad range of investigation and includes more traditional data-based activity, as well as literature reviews and historical research. Transmitting the results of one's scholarship through a printed medium requires development of formal, written communication skills. Promoting the refinement of such skills is another goal for the journal."
• Modern Psychological Studies "Modern Psychological Studies (MPS) is a psychological journal devoted exclusively to publishing manuscripts by undergraduate students. We are continuously seeking quality manuscripts for publication, and will consider manuscripts in any area of psychology. Although MPS primarily focuses on results from experimental research, there are also publication opportunities for theoretical papers, literature reviews and book reviews."
• PSI CHI Journal of Psychological Research "The Psi Chi Journal of Psychological Research encourages all Psi Chi members—undergraduate students, graduate students, and faculty—to submit manuscripts for publication. Submissions are accepted for review throughout the year. Although manuscripts are limited to empirical research, they may cover any topical area in the psychological sciences."
• The Yale Review of Undergraduate Research in Psychology "... is an annual journal that showcases the best and most original research in psychology conducted by undergraduates from around the world. We publish research in all areas of psychology, including clinical, developmental, cognitive, and social psychology. Our goal is to contribute to the scientific advance by encouraging serious, quality research early in students' academic careers. We provide a platform for undergraduate scientists to share their findings, and aim to bring together a community of young psychologists from both the United States and abroad."
• Undergraduate Research Journal for the Human Sciences ". . . is an annual online national, reviewed journal dedicated to the publication of undergraduate student research, with the twofold purpose of fostering and rewarding the scholarly efforts of undergraduate human sciences students as well as providing a valuable learning experience."

Subject Guide

• << Previous: Research Management Systems
• Last Updated: Dec 1, 2023 2:40 PM
• URL: https://libguides.grace.edu/JIBS

• Open access
• Published: 24 November 2023

Social media use and everyday cognitive failure: investigating the fear of missing out and social networks use disorder relationship

• Christian Montag 1 &
• Sebastian Markett 2  

BMC Psychiatry volume  23 , Article number:  872 ( 2023 ) Cite this article

417 Accesses

2 Altmetric

Metrics details

Nearly five billion individuals worldwide are using social media platforms. While the benefits of using social media, such as fostering social connections, are clear, ongoing discussions are focused on whether excessive use of these platforms might have adverse effects on cognitive functioning. Excessive social media use shares similarities with addictive behaviors and is believed to result from a complex interplay of individual characteristics, emotions, thoughts, and actions. Among these contributing factors, one of particular interest is the Fear of Missing Out (FoMO), a state where an individual apprehends that others are experiencing rewarding moments in their absence (but see more information on the FoMO trait/state debate in this article).

In this study, we aimed to explore the intricate relationships between FoMO, tendencies towards Social Networks Use Disorder (SNUD), and everyday cognitive failures. To achieve this, we gathered a large sample of N = 5314 participants and administered a comprehensive set of questionnaires. These included a Fear of Missing Out (FoMO) scale, which assessed both trait and state aspects of FoMO, the Social Networking Sites-Addiction Test (SNS-AT), designed to gauge tendencies towards SNUD, and the Cognitive Failure Questionnaire (CFQ), which measured everyday cognitive lapses.

Our findings revealed that among non-users of social media, both FoMO and everyday cognitive failures were at their lowest levels. Further, in the group of social media users, we observed a significant relationship between FoMO and cognitive failures, which was mediated by SNUD tendencies. This mediation was particularly pronounced for the state component of FoMO, which encompasses maladaptive thoughts related to online behavior.

Conclusions

While our study is cross-sectional and thus cannot establish causality, one plausible interpretation of our findings is that higher FoMO tendencies may trigger excessive social media use, which in turn could lead to cognitive failures, possibly due to distraction and reduced attention to everyday tasks.

Peer Review reports

In 2023, it’s estimated that nearly five billion people worldwide will be using social media, underscoring its global significance. Social media platforms hold great appeal for users due to their capacity to facilitate the establishment of social connections and the cultivation of social capital [ 1 ]. However, beyond the advantages of social media, there is an ongoing debate surrounding the platforms’ business model that relies on user data as a form of payment and the attention economy. This model has raised concerns about adverse consequences, including the potential for increased time spent online, which may lead to excessive use [ 2 ] (sometimes referred to as “social media addiction”), privacy breaches, and the dissemination of misinformation campaigns [ 3 ].

The focus of our present study centers on the issue of excessive social media. Within the scientific community, there is an active discourse regarding the precise nature of excessive social media consumption. One aspect under discussion is whether overuse of social media should be classified as an addictive behavior. This debate remains unsettled at this time [ 4 , 5 ]. In alignment with the nomenclature established in the context of Gaming Disorder, which describes addictive behaviors related to video gaming in the ICD-11 [ 6 ], our study employs the term “social networks use disorder” (SNUD) to characterize excessive social media use [ 7 , 8 ]. Others in the field may currently prefer the term “problematic social media use”, for further discussions around labeling see a work by Elhai, Yang and Levine [ 9 ]. It’s important to note that in our study, we prefer the term SNUD tendencies. This said, we conducted our research using a subclinical sample and emphasize that we do not intend to pathologize everyday behavior by employing the term SNUD [ 10 ].

In understanding the progression towards SNUD, the I-PACE model proves to be a valuable framework [ 11 ]. This model illustrates that Internet Use Disorders, including SNUD, result from the interaction of person, affect, cognition, and execution variables. One crucial variable shedding light on SNUD is the Fear of Missing Out, commonly abbreviated as FoMO [ 12 ]. From our view FoMO could be seen as a cognitive, but also affective state wherein an individual fears that others are having rewarding experiences in their absence; but see this work [ 13 ]. It is worth noting that FoMO exhibits correlations with traits such as neuroticism and low conscientiousness [ 14 ], blurring the line between a trait-like dimension and a state. Hence, depending on the perspective, FoMO could be seen as a trait or state. In this context, Wegmann et al. offer an intriguing perspective: They have developed a modified FoMO scale that provides insights into both trait FoMO and state FoMO [ 15 ]. Here, ‘trait’ refers to experiencing FoMO in general, not limited to online environments like social media. Conversely, the items designed to assess state FoMO specifically address the FoMO in online realms, such as constantly being online to avoid missing something. Understanding to what extent trait and state FoMO are differently related to the present variables of interest (SNUD and Cognitive Failure Questionnaire - CFQ; see below) will help other researchers to understand what FoMO variables to best choose in their studies. In our current study, we theoretically position the FoMO trait/state scale within the ‘P-variable’ of the I-PACE model. As mentioned above, the P-variable stands for “person” and comprises among others personality traits.

In addition to assessing FoMO, we also examined individual differences in everyday cognitive failure in the present study [ 16 ]. Cognitive failures refer to minor lapses in thought and action, such as forgetting appointments, overlooking information, or accidentally knocking things over [ 17 ]. These lapses are a natural consequence of fluctuations in various cognitive domains, including attention, memory, and action control [ 18 ]. The susceptibility to cognitive failure derives from a blend of both stable personality factors and situational elements [ 19 ]. The latter may encompass conditions such as sleep deprivation, stress, boredom, and information overload [ 20 , 21 , 22 , 23 ]. Recent research has also indicated that excessive use of social media could potentially instigate cognitive failure [ 24 ]. This observation concurs with the hypothesis that the incessant distractions emanating from smartphones and social media platforms might contribute to decreased productivity [ 25 , 26 ]. Consistent with this, a recent study utilizing a within-study design found that smartphone-checking behavior was correlated with a higher degree of cognitive failure [ 27 ]. It’s worth noting that even inverse associations were observed in this study when examining the screen-time measure for social media and other tool applications. This underscores the complexity of the relationships between objective smartphone usage measures and cognitive failure.

Returning to the topic of interruptions resulting from technology use within the context of investigating cognitive failure: Frequent interruptions triggered by incoming push notifications, which can also incite FoMO [ 28 ], could likewise contribute to cognitive lapses and reduced productivity. Our anticipation was that higher levels of FoMO, particularly online FoMO (referred to as state FoMO), would correlate with an increased occurrence of cognitive failures. We hypothesized that this relationship would be mediated by tendencies towards SNUD. Put differently, we expected that heightened levels of (state) FoMO might lead to elevated SNUD tendencies, which could, in turn, result in more frequent cognitive failures. This hypothesis is in line with prior research examining the connection between SNUD tendencies and cognitive failure [ 29 ]. It’s important to mention that another study also supports the exploration of the link between SNUD and cognitive failure. Hadlington [ 30 ] already observed that excessive mobile phone use, which is correlated with SNUD tendencies [ 31 ], was positively associated with self-reported cognitive failure. See also another study linking mobile phone addiction to cognitive failures [ 32 ].

We recruited a total of 5,530 participants through an online website, with the primary aim of investigating the relationships between cognitive failures, FoMO, and SNUD tendencies (other research questions from this data set will be investigated in the future; such as on TikTok Use Disorder and personality) Footnote 1 . The study was promoted through a series of media appearances, including print and radio, with a specific emphasis on its investigation of everyday cognitive failures. In the context of this study, participants completed questionnaires designed to assess their levels of FoMO, SNUD tendencies, and everyday cognitive failure. Detailed descriptions of these questionnaires will be provided in the following sections. As a token of appreciation for their participation, participants received insights into their own cognitive failure scores compared to those of other participants. The study was approved by the local ethics committee at Humboldt University in Berlin, Germany.

Data cleaning

From the initial N = 5,530 participants, we excluded a total of n = 25 individuals who identified as a third gender, as their numbers were insufficient for meaningful statistical analysis. Additionally, n = 3 participants were excluded because their questionnaire responses showed no variance. Moreover, n = 11 participants who were under 18 years old (the study only foresaw to include persons of 18 years and older), and n = 4 participants whose reported age fell outside of the defined upper age cutoff (set at 1.5 times the interquartile range over the third quartile of the age distribution) were also excluded from the study. Furthermore, due to the later introduction of the Fear of Missing Out (FoMO) scale, n = 173 participants were excluded because they did not provide FoMO scores. The final sample comprised n = 5314 participants (1801 males, 3513 females; mean-age: 53.43, SD = 14.84, range 18–94 years). The wider age range is attributable to media outlets where interviews were conducted. Please note that the entire study was conducted in the German language. Regrettably, we did not inquire about participants’ proficiency in the German language, and as a result, we were unable to screen out individuals who may have had difficulty comprehending the items in the online survey. However, we would not assume that participants navigated through the study with the sole intent of receiving feedback on their scores without understanding the content.

Questionnaires

Cognitive failure questionnaire.

Participants completed the Cognitive Failure Questionnaire (CFQ) first [ 16 ]. The CFQ, in its German version, consists of 32 items [ 33 ]. Participants rated the frequency with which various cognitive failures occurred to them in the past six months on a scale ranging from 0 (never) to 4 (very often). The questionnaire demonstrated excellent internal consistencies (α = 0.924, ω = 0.926), with higher scores indicating greater cognitive failure tendencies.

Next, participants filled in Wegmann’s FoMO scale [ 15 ], which includes both trait and state facets of Fear of Missing Out (FoMO). This German version of the scale is an adaptation of the original FoMO scale [ 13 ]. Wegmann’s FoMO scale comprises twelve items, with five items assessing trait FoMO and seven items examining state FoMO. Participants provided their responses on a five-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree). The internal consistencies for this scale were good (α = 0.788 and ω = 0.810 for trait FoMO; and α = 0.779 and ω = 0.792 for state FoMO), and higher scores indicated greater trait or state FoMO.

Social networking sites-addiction test

Finally, participants completed a modified version of the Bergen Facebook Addiction Scale (BFAS)/Bergen Social Media Addiction Scale (BSMAS) [ 34 , 35 ] as presented in Montag et al. [ 36 ] called Social Networking Sites-Addiction Test (SNS-AT): Unlike the BFAS, this version focuses on general social media overuse (not limited to Facebook, as with the BSMAS further developed from the BFAS [ 35 ]) and formulates items in the first person perspective (which is different than the BSMAS). The scale consists of six items, and participants rated them on a five-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree). The internal consistencies for this scale were also excellent (α = 0.864, ω = 0.868), with higher scores indicating greater tendencies toward SNUD.

Additionally, all participants were further asked if they used social media or not.

Statistical analyses

Data cleaning and visualization was performed in MATLAB (v2022b). Statistical analyses were computed with the Jamovi package 2.3.18.0. Participants were categorized into one of three groups based on their social media usage patterns. The first group comprised all participants who indicated that they were active social media users (this means they stated to use social media). The second group consisted of participants who declared that they did not use social media and also exhibited no SNUD tendencies, as evidenced by obtaining the minimum score on the SNS-AT. The third group comprised all participants who stated that they did not use social media but reported SNUD tendencies on the SNS-AT (scores > 6). This peculiar combination of not using social media while still displaying tendencies toward problematic use could imply temporary abstinence or an intentional avoidance of social media due to perceived negative consequences. However, it’s also plausible that individuals in this group provided inconsistent responses for other reasons. Regrettably, we did not include items about past social media usage, making it impossible to verify the status of “ex-users.“ Despite this limitation, we chose to compare this group with the other two for exploratory purposes, but we advise interpreting the results with appropriate caution.

Descriptive statistics for FoMO, SNUD tendencies, and cognitive failure in the three groups are presented in the main text, detailed descriptive statistics for male and female subsamples are given in the supplementary material. The three groups were contrasted by MANOVA with the FoMO, SNS-AT and CFQ as dependent variables and gender and ages as covariates. This was done due to varying age and gender ratios in the three groups (see also supplementary material).

To test our hypothesis regarding the potential mediation of the relationship between FoMO and cognitive failure via SNUD tendencies, we followed a structured approach. Initially, we examined the pairwise relationships between the variables using linear correlation analyses. Subsequently, we employed Jamovi’s advanced mediation model module to construct mediation models. In these models, we designated FoMO as the predictor, SNUD tendencies as the mediating variable, and cognitive failure as the outcome variable. This terminology aligns with mediation model conventions and should not be interpreted as implying causality. We conducted separate mediation models for state and trait FoMO.

Open science and transparency statement

We collected additional measures from the participants for other research questions. The data linked to the present report are available on the Open Science Framework, together with the analysis code for data cleaning ( https://osf.io/bd9y4/ ).

Descriptive statistics are presented in Table  1 . The three groups differed in all dependent measures (see Fig.  1 ; Table  2 ). Social media users (n = 3618; 1147 males, 2471 females, mean-age: 50,89, SD = 14,81) reported higher state and trait FoMO, more SNUD tendencies on the SNS-AT, and more frequent cognitive failure than non-users of social media who did not report SNUD tendencies (n = 1148; 439 males, 709 females; mean-age: 59,47, SD = 12,75). Non-users of social media who still reported higher SNUD tendencies had also higher scores on all measures compared to the non-user group without SNUD tendencies (n = 548; 215 males, 333 females; mean-age: 57,51, SD = 14,53). The scores in this group resembled the scores of active social media users (see Fig.  1 ). More detailed descriptive statistics regarding males and females are presented in the Supplementary Table  1 .

Descriptive statistics (means and standard errors) for trait and state FoMO, SNUD tendencies, and cognitive failure in the three study groups. Results are plotted for male and female participants separately, as indicated by different colors of the lines

Focusing only on the group of active social media users (n = 3618), we observed significant intercorrelations between all study variables (see Table  3 ). Correlation coefficients between the questionnaire measures indicated moderate effect sizes (exception: CFQ and state FoMO was in a small effect size area). Age was inversely related to all questionnaire variables with small to moderate effect sizes.

The subsequent mediation analysis revealed a complete mediation of the relationship between state FoMO and cognitive failure through SNUD tendencies and a partial mediation of the relationship between trait FoMO and cognitive failure through SNUD tendencies (see Fig.  2 ). Adding age and gender as additional factors to the model did not change these mediations in a meaningful way (see supplementary materials).

Mediation models for trait FoMO (left side: A ) and state FoMO (right side: B ). Effect estimates with standard errors are given for the indirect effect, its two components, the direct effect, and the total effect

The primary objective of this study was to explore the intricate relationships between FoMO, SNUD tendencies, and cognitive failures. Our findings bring these associations to the forefront, providing insight into the complex interconnections between these factors. Significantly, this study offers a comprehensive examination, encompassing both trait and state FoMO, SNUD tendencies, and cognitive failures as assessed by the Cognitive Failure Questionnaire (CFQ) within a single investigation.

In the group of active social media users, we identified a network of positive associations among all three variables. As depicted in Table  3 the overall correlations between FoMO, SNUD and CFQ are in the moderate effect size area (with the exception for the FoMO state-CFQ association, which falls in the small effect size area). Therefore, one can assume robust associations between the variables, at least in the active social media user group. These findings align with prior research, which consistently shows robust connections between FoMO and elevated SNUD tendencies [ 12 ]. This further underscores the notion that the FoMO on online experiences is closely linked to the inclination toward problematic social media use (or SNUD tendencies). The present study also highlights the importance of differentiating between the trait and state facets of FoMO variables. To elaborate, we found that the trait FoMO variable exhibited a stronger association with CFQ than the state FoMO variable did. Conversely, when examining FoMO and SNUD tendencies, we observed the opposite pattern. Notably, state FoMO (specifically, online FoMO) showed a stronger association with SNUD tendencies than the trait FoMO did. This distinction arises because the trait FoMO items capture a more general sense of FoMO without explicitly referencing the online context in their respective items.

The examination of connections between SNUD tendencies and cognitive failures remains a relatively uncharted area. However, our findings are consistent with emerging evidence. For instance, a study observed associations between performance indices on cognitive tasks, such as the Wisconsin Card Sorting Test, and scores on the Bergen Social Media Addiction Scale (BSMAS) in problematic social media users [ 37 ]. Similarly, other studies have linked “dependence on SNS” (p. 121) and more cognitive failures [ 29 ] with consistent findings reported in a bit older work dealing with problematic mobile phone use and cognitive failure [ 30 ], see also a more recent study [ 38 ]. Niu et al. also noted that smartphone presence can adversely affect cognitive functions, with FoMO serving as a moderating variable [ 39 ].

To the best of our knowledge, our study stands as one of the first to comprehensively investigate FoMO (including both trait and state scales), SNUD tendencies, and cognitive failure (as assessed by the CFQ) within a single study. Our results corroborate the hypothesized positive associations between FoMO and cognitive failure, with SNUD tendencies emerging as a mediator in this relationship. Specifically, we found that SNUD tendencies fully mediated the relationship between state FoMO and cognitive failure, while the relationship between trait FoMO and cognitive failure exhibited a partial mediation through SNUD tendencies. The slight differences in the mediation models likely can be explained by the different strength of associations as mentioned earlier in the discussion.

While our findings provide valuable insights, several limitations warrant consideration. Firstly, our study’s cross-sectional nature precludes any inference of causality. In principle, it is also imaginable that persons scoring higher on cognitive failure are more prone to experience FoMO and SNUD. Perhaps persons with more cognitive failure also have more problems in self-regulation and develop more easily SNUD tendencies. To establish causal relationships, future experimental research is needed, exploring diverse variables within our mediation model. Secondly, self-report measures inherently carry the potential for biases, including social desirability and a lack of introspection among some participants. Additionally, our non-representative sample limits the generalizability of our findings, although they are broadly consistent with existing literature. Finally, other variables are of interest to be studied in this context of the present research question including personality traits or other sociodemographic variables. This would make an interesting research endeavor.

Despite these limitations, our study underscores the robust associations among FoMO, SNUD tendencies, and cognitive failure. The possibility that excessive social media use may lead to cognitive failures highlights the importance of designing healthier social media platforms. Platforms that focus on user well-being rather than exploiting addictive design elements can contribute to a more sustainable and responsible digital landscape [ 40 , 41 ]. Such efforts may ultimately require a shift away from data-driven business models, encouraging exploration of alternative payment structures in the world of social media [ 42 ].

In conclusions, our study contributes to the growing body of knowledge about the intricate relationships between FoMO, SNUD tendencies, and cognitive failures. It underscores the need for further research and proactive measures to promote healthier and more mindful engagement with social media platforms in our digitally connected world.

Data availability

The data is available via this link at the Open Science Framework: https://osf.io/bd9y4/ .

Of note, only few participants from the present sample stated to use TikTok. Therefore, this data cannot be presented here and the data collection in this regard is ongoing. Please note that the study was not framed to investigate TikTok use and so no bias in data gathering in this regard could have happened.

Ellison N, Steinfield C, Lampe C. The benefits of Facebook friends: Social Capital and College Students’ Use of Online Social Network sites. J Computer-Mediated Communication. 2007;12:1143–68. https://doi.org/10.1111/j.1083-6101.2007.00367.x .

Article   Google Scholar  

Montag C, Elhai JD. On Social Media Design, (Online-)Time Well-spent and Addictive Behaviors in the Age of Surveillance Capitalism. Curr Addict Rep [Internet]. 2023 [cited 2023 Jun 28]; https://doi.org/10.1007/s40429-023-00494-3 .

Montag C, Hegelich S. Understanding Detrimental Aspects of Social Media Use: Will the Real Culprits Please Stand Up? Frontiers in Sociology [Internet]. 2020 [cited 2023 Jul 10];5. Available from: https://www.frontiersin.org/articles/ https://doi.org/10.3389/fsoc.2020.599270 .

Brand M, Rumpf H-J, Demetrovics Z, MÜller A, Stark R, King DL et al. Which conditions should be considered as disorders in the International Classification of Diseases (ICD-11) designation of “other specified disorders due to addictive behaviors”? Journal of Behavioral Addictions [Internet]. 2020 [cited 2021 Sep 6];1. Available from: https://doi.org/10.1556/2006.2020.00035 .

Carbonell X, Panova T. A critical consideration of social networking sites’ addiction potential. Addiction Research & Theory [Internet]. 2017 [cited 2020 Aug 24];25:48–57. https://doi.org/10.1080/16066359.2016.1197915 .

Montag C, Schivinski B, Pontes H. Is the proposed distinction of Gaming Disorder into a predominantly online vs. offline form meaningful? Empirical evidence from a large German speaking gamer sample. Addict Behav Rep. 2021;14:100391. https://doi.org/10.1016/j.abrep.2021.100391 .

PubMed   PubMed Central   Google Scholar  

Montag C, Wegmann E, Sariyska R, Demetrovics Z, Brand M. How to overcome taxonomical problems in the study of Internet use disorders and what to do with “smartphone addiction”? Journal of Behavioral Addictions [Internet]. 2021 [cited 2021 Mar 6];9:908–14. Available from: https://akjournals.com/view/journals/2006/9/4/article-p908.xml .

Rumpf H-J, Batra A, Bischof A, Hoch E, Lindenberg K, Mann K et al. Vereinheitlichung der Bezeichnungen für Verhaltenssüchte. SUCHT [Internet]. 2021 [cited 2021 Nov 22];67:181–5. Available from: https://econtent.hogrefe.com/doi/full/ https://doi.org/10.1024/0939-5911/a000720 .

Elhai JD, Yang H, Levine JC. Applying fairness in labeling various types of internet use disorders. •: Commentary on How to overcome taxonomical problems in the study of internet use disorders and what to do with “smartphone addiction”? Journal of Behavioral Addictions [Internet]. 2020 [cited 2022 Dec 8];9:924–7. Available from: https://akjournals.com/view/journals/2006/9/4/article-p924.xml .

Billieux J, Schimmenti A, Khazaal Y, Maurage P, Heeren A. Are we overpathologizing everyday life? A tenable blueprint for behavioral addiction research. Journal of Behavioral Addictions [Internet]. 2015 [cited 2020 Oct 13];4:119–23. Available from: https://akjournals.com/view/journals/2006/4/3/article-p119.xml .

Brand M, Wegmann E, Stark R, Müller A, Wölfling K, Robbins TW et al. The Interaction of Person-Affect-Cognition-Execution (I-PACE) model for addictive behaviors: Update, generalization to addictive behaviors beyond internet-use disorders, and specification of the process character of addictive behaviors. Neuroscience & Biobehavioral Reviews [Internet]. 2019 [cited 2020 Mar 31];104:1–10. Available from: http://www.sciencedirect.com/science/article/pii/S0149763419303707 .

Elhai JD, Yang H, Montag C. Fear of missing out (FOMO): overview, theoretical underpinnings, and literature review on relations with severity of negative affectivity and problematic technology use. Braz J Psychiatry. 2021;43:203–9. https://doi.org/10.1590/1516-4446-2020-0870 .

Article   PubMed   Google Scholar  

Przybylski AK, Murayama K, DeHaan CR, Gladwell V. Motivational, emotional, and behavioral correlates of fear of missing out. Computers in Human Behavior [Internet]. 2013 [cited 2021 Dec 21];29:1841–8. Available from: https://www.sciencedirect.com/science/article/pii/S0747563213000800 .

Rozgonjuk D, Sindermann C, Elhai JD, Montag C. Individual differences in Fear of Missing Out (FoMO): Age, gender, and the Big Five personality trait domains, facets, and items. Personality and Individual Differences [Internet]. 2021 [cited 2021 Nov 22];171:110546. Available from: https://www.sciencedirect.com/science/article/pii/S0191886920307376 .

Wegmann E, Oberst U, Stodt B, Brand M. Online-specific fear of missing out and Internet-use expectancies contribute to symptoms of Internet-communication disorder. Addictive Behaviors Reports [Internet]. 2017 [cited 2021 Aug 19];5:33–42. Available from: https://www.sciencedirect.com/science/article/pii/S235285321730007X .

Broadbent DE, Cooper PF, FitzGerald P, Parkes KR. The cognitive failures questionnaire (CFQ) and its correlates. Br J Clin Psychol. 1982;21:1–16. https://doi.org/10.1111/j.2044-8260.1982.tb01421.x .

Article   CAS   PubMed   Google Scholar  

Wagle AC, Berrios GE, Ho L. The cognitive failures questionnaire in psychiatry. Comprehensive Psychiatry [Internet]. 1999 [cited 2023 Oct 11];40:478–84. Available from: https://www.sciencedirect.com/science/article/pii/S0010440X99900937 .

Carrigan N, Barkus E. A systematic review of cognitive failures in daily life: healthy populations. Neurosci Biobehavioral Reviews. 2016;63:29–42. https://doi.org/10.1016/j.neubiorev.2016.01.010 .

Markett S, Reuter M, Sindermann C, Montag C. Cognitive failure susceptibility and personality: Self-directedness predicts everyday cognitive failure. Personality and Individual Differences [Internet]. 2020 [cited 2023 Aug 10];159:109916. Available from: https://www.sciencedirect.com/science/article/pii/S0191886920301057 .

Willert MV, Thulstrup AM, Hertz J, Bonde JP. Sleep and cognitive failures improved by a three-month stress management intervention. Int J Stress Manage. 2010;17:193–213. https://doi.org/10.1037/a0019612 .

Wallace C, Kass S, Stanny C. The cognitive failures Questionnaire Revisited: dimensions and correlates. J Gen Psychol. 2002;129:238–56. https://doi.org/10.1080/00221300209602098 .

Wallace JC, Vodanovich SJ, Restino BM. Predicting cognitive failures from boredom proneness and daytime sleepiness scores: an investigation within military and undergraduate samples. Pers Indiv Differ. 2003;34:635–44. https://doi.org/10.1016/S0191-8869(02)00050-8 .

Whelan E, Islam N, Brooks S. Cognitive control and social media overload. Proc Am Conf Info Syst. 2017;12. https://aisel.aisnet.org/amcis2017/SocialComputing/Presentations/12 .

Sharifian N, Zahodne LB. Daily associations between social media use and memory failures: the mediating role of negative affect. J Gen Psychol. 2021;148:67–83. https://doi.org/10.1080/00221309.2020.1743228 .

Duke É, Montag C. Smartphone addiction, daily interruptions and self-reported productivity. Addictive Behaviors Reports [Internet]. 2017 [cited 2020 Mar 31];6:90–5. Available from: http://www.sciencedirect.com/science/article/pii/S2352853217300159 .

Rozgonjuk D, Sindermann C, Elhai JD, Montag C. Fear of Missing Out (FoMO) and social media’s impact on daily-life and productivity at work: Do WhatsApp, Facebook, Instagram, and Snapchat Use Disorders mediate that association. Addict Behav [Internet]. 2020 [cited 2021 Nov 22];110:106487. https://doi.org/10.1016/j.addbeh.2020.106487 .

Hartanto A, Lee KYX, Chua YJ, Quek FYX, Majeed NM. Smartphone use and daily cognitive failures: A critical examination using a daily diary approach with objective smartphone measures. British Journal of Psychology [Internet]. 2023 [cited 2023 Oct 11];114:70–85. Available from: https://onlinelibrary.wiley.com/doi/abs/ https://doi.org/10.1111/bjop.12597 .

Alutaybi A, Arden-Close E, McAlaney J, Stefanidis A, Phalp K, Ali R. How Can Social Networks Design Trigger Fear of Missing Out? 2019 IEEE International Conference on Systems, Man and Cybernetics (SMC). 2019;3758–65. https://doi.org/10.1109/SMC.2019.8914672 .

Xanidis N, Brignell CM. The association between the use of social network sites, sleep quality and cognitive function during the day. Computers in Human Behavior [Internet]. 2016 [cited 2023 Aug 10];55:121–6. Available from: https://www.sciencedirect.com/science/article/pii/S0747563215301357 .

Hadlington LJ. Cognitive failures in daily life: Exploring the link with Internet addiction and problematic mobile phone use. Computers in Human Behavior [Internet]. 2015 [cited 2023 Aug 10];51:75–81. Available from: https://www.sciencedirect.com/science/article/pii/S0747563215003313 .

Rozgonjuk D, Sindermann C, Elhai JD, Christensen AP, Montag C. Associations between symptoms of problematic smartphone, Facebook, WhatsApp, and Instagram use: An item-level exploratory graph analysis perspective. Journal of Behavioral Addictions [Internet]. 2020 [cited 2021 Jan 24];9:686–97. Available from: https://akjournals.com/view/journals/2006/9/3/article-p686.xml .

Hong W, Liu R-D, Ding Y, Sheng X, Zhen R. Mobile phone addiction and cognitive failures in daily life: The mediating roles of sleep duration and quality and the moderating role of trait self-regulation. Addictive Behaviors [Internet]. 2020 [cited 2023 Aug 10];107:106383. Available from: https://www.sciencedirect.com/science/article/pii/S0306460319311839 .

Klumb PL. Cognitive failures and performance differences: validation studies of a German version of the cognitive failures questionnaire. Ergonomics. 1995;38:1456–67. https://doi.org/10.1080/00140139508925202 .

Andreassen CS, Torsheim T, Brunborg GS, Pallesen S. Development of a Facebook Addiction Scale. Psychol Rep [Internet]. 2012 [cited 2023 May 13];110:501–17. Available from: http://journals.sagepub.com/doi/ https://doi.org/10.2466/02.09.18.PR0.110.2.501-517 .

Andreassen CS, Billieux J, Griffiths MD, Kuss DJ, Demetrovics Z, Mazzoni E, et al. The relationship between addictive use of social media and video games and symptoms of psychiatric disorders: a large-scale cross-sectional study. Psychol Addict Behav. 2016;30:252–62. https://doi.org/10.1037/adb0000160 .

Montag C, Müller M, Pontes HM, Elhai JD. On fear of missing out, social networks use disorder tendencies and meaning in life. BMC Psychology [Internet]. 2023 [cited 2023 Oct 27];11:358. https://doi.org/10.1186/s40359-023-01342-9 .

Aydın O, Obuća F, Boz C, Ünal-Aydın P. Associations between executive functions and problematic social networking sites use. Journal of Clinical and Experimental Neuropsychology [Internet]. 2020 [cited 2023 Aug 10];42:634–45. https://doi.org/10.1080/13803395.2020.1798358 .

Zhang B, Peng Y, Luo X, Mao H, Luo Y, Hu R et al. Mobile phone addiction and cognitive failures in Chinese adolescents: The role of rumination and mindfulness. Journal of Psychology in Africa [Internet]. 2021 [cited 2023 Aug 10];31:49–55. https://doi.org/10.1080/14330237.2020.1871239 .

Niu G, Shi X, Zhang Z, Yang W, Jin S, Sun X. Can smartphone presence affect cognitive function? The moderating role of fear of missing out. Computers in Human Behavior [Internet]. 2022 [cited 2023 Aug 10];136:107399. Available from: https://www.sciencedirect.com/science/article/pii/S0747563222002217 .

Flayelle M, Brevers D, King DL, Maurage P, Perales JC, Billieux J. A taxonomy of technology design features that promote potentially addictive online behaviours. Nat Rev Psychol [Internet]. 2023 [cited 2023 Mar 7];2:136–50. Available from: https://www.nature.com/articles/s44159-023-00153-4 .

Montag C, Lachmann B, Herrlich M, Zweig K. Addictive Features of Social Media/Messenger Platforms and Freemium Games against the Background of Psychological and Economic Theories. International Journal of Environmental Research and Public Health [Internet]. 2019 [cited 2020 Mar 31];16:2612. Available from: https://doi.org/10.3390/ijerph16142612 .

Dhawan S, Hegelich S, Sindermann C, Montag C. Re-start social media, but how? Telematics and Informatics Reports [Internet]. 2022 [cited 2023 Jul 10];8:100017. Available from: https://www.sciencedirect.com/science/article/pii/S2772503022000159 .

Download references

Acknowledgements

Open Access funding enabled and organized by Projekt DEAL. Sebastian Markett is supported by Open Humboldt Freiräume with funds from the Berlin University Alliance within the Excellence Strategy of the German Federal and State Governments.

Author information

Authors and affiliations.

Department of Molecular Psychology, Institute of Psychology and Education, Ulm University, Helmholtzstr. 8/1, 89081, Ulm, Germany

Christian Montag

Molecular Psychology, Department of Psychology, Humboldt-Universität zu Berlin, Berlin, Germany

Sebastian Markett

You can also search for this author in PubMed   Google Scholar

Contributions

CM and SM designed the present study, collected the data and analyzed the data. Christian Montag wrote the first complete draft of the paper, which was revised by SM. Both authors read and approved the final manuscript.

Corresponding author

Correspondence to Christian Montag .

Ethics declarations

Ethics approval and consent to participate.

The study was approved by the local ethics committee at Humboldt University in Berlin, Germany and all participants provided informed e-consent.

Consent for publication

Does not apply.

Competing interests

Dr. Montag reports no conflict of interest. However, for reasons of transparency Dr. Montag mentions that he has received (to Ulm University and earlier University of Bonn) grants from agencies such as the German Research Foundation (DFG). Dr. Montag has performed grant reviews for several agencies; has edited journal sections and articles; has given academic lectures in clinical or scientific venues or companies; and has generated books or book chapters for publishers of mental health texts. For some of these activities he received royalties, but never from gaming or social media companies. Dr. Montag mentions that he was part of a discussion circle (Digitalität und Verantwortung: https://about.fb.com/de/news/h/gespraechskreis-digitalitaet-und-verantwortung/ ) debating ethical questions linked to social media, digitalization and society/democracy at Facebook. In this context, he received no salary for his activities. Finally, he mentions that he currently functions as independent scientist on the scientific advisory board of the Nymphenburg group (Munich, Germany). This activity is financially compensated. Moreover, he is on the scientific advisory board of Applied Cognition (Redwood City, CA, USA), an activity which is also compensated. Dr. Markett reports no conflict of interest.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Cite this article.

Montag, C., Markett, S. Social media use and everyday cognitive failure: investigating the fear of missing out and social networks use disorder relationship. BMC Psychiatry 23 , 872 (2023). https://doi.org/10.1186/s12888-023-05371-x

Download citation

Received : 24 August 2023

Accepted : 10 November 2023

Published : 24 November 2023

DOI : https://doi.org/10.1186/s12888-023-05371-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Social media addiction
• Social networks use disorder
• Fear of missing out
• Cognitive failure

BMC Psychiatry

ISSN: 1471-244X

• Submission enquiries: [email protected]
• General enquiries: [email protected]

• UNO Criss Library

Social Work Research Guide

What is a research journal.

• Find Articles
• Find E-Books and Books

Reading an Academic Article

• Free Online Resources
• Reference and Writing
• Citation Help This link opens in a new window

Anatomy of a Scholarly Article

TIP: When possible, keep your research question(s) in mind when reading scholarly articles. It will help you to focus your reading.

Title : Generally are straightforward and describe what the article is about. Titles often include relevant key words.

Abstract : A summary of the author(s)'s research findings and tells what to expect when you read the full article. It is often a good idea to read the abstract first, in order to determine if you should even bother reading the whole article.

Discussion and Conclusion : Read these after the Abstract (even though they come at the end of the article). These sections can help you see if this article will meet your research needs. If you don’t think that it will, set it aside.

Introduction : Describes the topic or problem researched. The authors will present the thesis of their argument or the goal of their research.

Literature Review : May be included in the introduction or as its own separate section. Here you see where the author(s) enter the conversation on this topic. That is to say, what related research has come before, and how do they hope to advance the discussion with their current research?

Methods : This section explains how the study worked. In this section, you often learn who and how many participated in the study and what they were asked to do. You will need to think critically about the methods and whether or not they make sense given the research question.

Results : Here you will often find numbers and tables. If you aren't an expert at statistics this section may be difficult to grasp. However you should attempt to understand if the results seem reasonable given the methods.

Works Cited (also be called References or Bibliography ): This section comprises the author(s)’s sources. Always be sure to scroll through them. Good research usually cites many different kinds of sources (books, journal articles, etc.). As you read the Works Cited page, be sure to look for sources that look like they will help you to answer your own research question.

Adapted from http://library.hunter.cuny.edu/research-toolkit/how-do-i-read-stuff/anatomy-of-a-scholarly-article

A research journal is a periodical that contains articles written by experts in a particular field of study who report the results of research in that field. The articles are intended to be read by other experts or students of the field, and they are typically much more sophisticated and advanced than the articles found in general magazines. This guide offers some tips to help distinguish scholarly journals from other periodicals.

CHARACTERISTICS OF RESEARCH JOURNALS

PURPOSE : Research journals communicate the results of research in the field of study covered by the journal. Research articles reflect a systematic and thorough study of a single topic, often involving experiments or surveys. Research journals may also publish review articles and book reviews that summarize the current state of knowledge on a topic.

APPEARANCE : Research journals lack the slick advertising, classified ads, coupons, etc., found in popular magazines. Articles are often printed one column to a page, as in books, and there are often graphs, tables, or charts referring to specific points in the articles.

AUTHORITY : Research articles are written by the person(s) who did the research being reported. When more than two authors are listed for a single article, the first author listed is often the primary researcher who coordinated or supervised the work done by the other authors. The most highly‑regarded scholarly journals are typically those sponsored by professional associations, such as the American Psychological Association or the American Chemical Society.

VALIDITY AND RELIABILITY : Articles submitted to research journals are evaluated by an editorial board and other experts before they are accepted for publication. This evaluation, called peer review, is designed to ensure that the articles published are based on solid research that meets the normal standards of the field of study covered by the journal. Professors sometimes use the term "refereed" to describe peer-reviewed journals.

WRITING STYLE : Articles in research journals usually contain an advanced vocabulary, since the authors use the technical language or jargon of their field of study. The authors assume that the reader already possesses a basic understanding of the field of study.

REFERENCES : The authors of research articles always indicate the sources of their information. These references are usually listed at the end of an article, but they may appear in the form of footnotes, endnotes, or a bibliography.

PERIODICALS THAT ARE NOT RESEARCH JOURNALS

POPULAR MAGAZINES : These are periodicals that one typically finds at grocery stores, airport newsstands, or bookstores at a shopping mall. Popular magazines are designed to appeal to a broad audience, and they usually contain relatively brief articles written in a readable, non‑technical language.

Examples include: Car and Driver , Cosmopolitan , Esquire , Essence , Gourmet , Life , People Weekly , Readers' Digest , Rolling Stone , Sports Illustrated , Vanity Fair , and Vogue .

NEWS MAGAZINES : These periodicals, which are usually issued weekly, provide information on topics of current interest, but their articles seldom have the depth or authority of scholarly articles.

Examples include: Newsweek , Time , U.S. News and World Report .

OPINION MAGAZINES : These periodicals contain articles aimed at an educated audience interested in keeping up with current events or ideas, especially those pertaining to topical issues. Very often their articles are written from a particular political, economic, or social point of view.

Examples include: Catholic World , Christianity Today , Commentary , Ms. , The Militant , Mother Jones , The Nation , National Review , The New Republic , The Progressive , and World Marxist Review .

TRADE MAGAZINES : People who need to keep up with developments in a particular industry or occupation read these magazines. Many trade magazines publish one or more special issues each year that focus on industry statistics, directory lists, or new product announcements.

Examples include: Beverage World , Progressive Grocer , Quick Frozen Foods International , Rubber World , Sales and Marketing Management , Skiing Trade News , and Stores .

Literature Reviews

• Literature Review Guide General information on how to organize and write a literature review.
• The Literature Review: A Few Tips On Conducting It Contains two sets of questions to help students review articles, and to review their own literature reviews.
• << Previous: Find E-Books and Books
• Next: Statistics >>
• Last Updated: Nov 21, 2023 7:48 AM
• URL: https://libguides.unomaha.edu/social_work

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 29 November 2023

Scaling deep learning for materials discovery

• Amil Merchant   ORCID: orcid.org/0000-0001-5262-6599 1   na1 ,
• Simon Batzner 1   na1 ,
• Samuel S. Schoenholz 1   na1 ,
• Muratahan Aykol   ORCID: orcid.org/0000-0001-6433-7217 1 ,
• Gowoon Cheon 2 &
• Ekin Dogus Cubuk   ORCID: orcid.org/0000-0003-0524-2837 1   na1  

Nature ( 2023 ) Cite this article

53k Accesses

1 Citations

546 Altmetric

Metrics details

• Computer science
• Scaling laws

Novel functional materials enable fundamental breakthroughs across technological applications from clean energy to information processing 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 . From microchips to batteries and photovoltaics, discovery of inorganic crystals has been bottlenecked by expensive trial-and-error approaches. Concurrently, deep-learning models for language, vision and biology have showcased emergent predictive capabilities with increasing data and computation 12 , 13 , 14 . Here we show that graph networks trained at scale can reach unprecedented levels of generalization, improving the efficiency of materials discovery by an order of magnitude. Building on 48,000 stable crystals identified in continuing studies 15 , 16 , 17 , improved efficiency enables the discovery of 2.2 million structures below the current convex hull, many of which escaped previous human chemical intuition. Our work represents an order-of-magnitude expansion in stable materials known to humanity. Stable discoveries that are on the final convex hull will be made available to screen for technological applications, as we demonstrate for layered materials and solid-electrolyte candidates. Of the stable structures, 736 have already been independently experimentally realized. The scale and diversity of hundreds of millions of first-principles calculations also unlock modelling capabilities for downstream applications, leading in particular to highly accurate and robust learned interatomic potentials that can be used in condensed-phase molecular-dynamics simulations and high-fidelity zero-shot prediction of ionic conductivity.

The discovery of energetically favourable inorganic crystals is of fundamental scientific and technological interest in solid-state chemistry. Experimental approaches over the decades have catalogued 20,000 computationally stable structures (out of a total of 200,000 entries) in the Inorganic Crystal Structure Database (ICSD) 15 , 18 . However, this strategy is impractical to scale owing to costs, throughput and synthesis complications 19 . Instead, computational approaches championed by the Materials Project (MP) 16 , the Open Quantum Materials Database (OQMD) 17 , AFLOWLIB 20 and NOMAD 21 have used first-principles calculations based on density functional theory (DFT) as approximations of physical energies. Combining ab initio calculations with simple substitutions has allowed researchers to improve to 48,000 computationally stable materials according to our own recalculations 22 , 23 , 24 (see Methods ). Although data-driven methods that aid in further materials discovery have been pursued, thus far, machine-learning techniques have been ineffective in estimating stability (decomposition energy) with respect to the convex hull of energies from competing phases 25 .

In this paper, we scale up machine learning for materials exploration through large-scale active learning, yielding the first models that accurately predict stability and, therefore, can guide materials discovery. Our approach relies on two pillars: first, we establish methods for generating diverse candidate structures, including new symmetry-aware partial substitutions (SAPS) and random structure search 26 . Second, we use state-of-the art graph neural networks (GNNs) that improve modelling of material properties given structure or composition. In a series of rounds, these graph networks for materials exploration (GNoME) are trained on available data and used to filter candidate structures. The energy of the filtered candidates is computed using DFT, both verifying model predictions and serving as a data flywheel to train more robust models on larger datasets in the next round of active learning.

Through this iterative procedure, GNoME models have discovered more than 2.2 million structures stable with respect to previous work, in particular agglomerated datasets encompassing computational and experimental structures 15 , 16 , 17 , 27 . Given that discovered materials compete for stability, the updated convex hull consists of 381,000 new entries for a total of 421,000 stable crystals, representing an-order-of-magnitude expansion from all previous discoveries. Consistent with observations in other domains of machine learning 28 , we observe that our neural networks predictions improve as a power law with the amount of data. Final GNoME models accurately predict energies to 11 meV atom −1 and improve the precision of stable predictions (hit rate) to above 80% with structure and 33% per 100 trials with composition only, compared with 1% in previous work 17 . Moreover, these networks develop emergent out-of-distribution generalization. For example, GNoME enables accurate predictions of structures with 5+ unique elements (despite omission from training), providing one of the first strategies to efficiently explore this chemical space. We validate findings by comparing predictions with experiments and higher-fidelity r 2 SCAN (ref.  29 ) computations.

Finally, we demonstrate that the dataset produced in GNoME discovery unlocks new modelling capabilities for downstream applications. The structures and relaxation trajectories present a large and diverse dataset to enable training of learned, equivariant interatomic potentials 30 , 31 with unprecedented accuracy and zero-shot generalization. We demonstrate the promise of these potentials for materials property prediction through the estimation of ionic conductivity from molecular-dynamics simulations.

Overview of generation and filtration

The space of possible materials is far too large to sample in an unbiased manner. Without a reliable model to cheaply approximate the energy of candidates, researchers guided searches by restricting generation with chemical intuition, accomplished by substituting similar ions or enumerating prototypes 22 . Although improving search efficiency 17 , 27 , this strategy fundamentally limited how diverse candidates could be. By guiding searches with neural networks, we are able to use diversified methods for generating candidates and perform a broader exploration of crystal space without sacrificing efficiency.

To generate and filter candidates, we use two frameworks, which are visualized in Fig. 1a . First, structural candidates are generated by modifications of available crystals. However, we strongly augment the set of substitutions by adjusting ionic substitution probabilities to give priority to discovery and use newly proposed symmetry aware partial substitutions (SAPS) to efficiently enable incomplete replacements 32 . This expansion results in more than 10 9 candidates over the course of active learning; the resulting structures are filtered by means of GNoME using volume-based test-time augmentation and uncertainty quantification through deep ensembles 33 . Finally, structures are clustered and polymorphs are ranked for evaluation with DFT (see Methods ). In the second framework, compositional models predict stability without structural information. Inputs are reduced chemical formulas. Generation by means of oxidation-state balancing is often too strict (for example, neglecting Li 15 Si 4 ). Using relaxed constraints (see Methods ), we filter compositions using GNoME and initialize 100 random structures for evaluation through ab initio random structure searching (AIRSS) 26 . In both frameworks, models provide a prediction of energy and a threshold is chosen on the basis of the relative stability (decomposition energy) with respect to competing phases. Evaluation is performed through DFT computations in the Vienna Ab initio Simulation Package (VASP) 34 and we measure both the number of stable materials discovered as well as the precision of predicted stable materials (hit rate) in comparison with the Materials Project 16 .

a , A summary of the GNoME-based discovery shows how model-based filtration and DFT serve as a data flywheel to improve predictions. b , Exploration enabled by GNoME has led to 381,000 new stable materials, almost an order of magnitude larger than previous work. c , 736 structures have been independently experimentally verified, with six examples shown 50 , 51 , 52 , 53 , 54 , 55 . d , Improvements from graph network predictions enable efficient discovery in combinatorial regions of materials, for example, with six unique elements, even though the training set stopped at four unique elements. e , GNoME showcases emergent generalization when tested on out-of-domain inputs from random structure search, indicating progress towards a universal energy model.

All GNoME models are GNNs that predict the total energy of a crystal. Inputs are converted to a graph through a one-hot embedding of the elements. We follow the message-passing formulation 35 , 36 , in which aggregate projections are shallow multilayer perceptrons (MLPs) with swish nonlinearities. For structural models, we find it important to normalize messages from edges to nodes by the average adjacency of atoms across the entire dataset. Initial models are trained on a snapshot of the Materials Project from 2018 of approximately 69,000 materials. Previous work benchmarked this task at a mean absolute error (MAE) of 28 meV atom −1 (ref.  37 ); however, we find that the improved networks achieve a MAE of 21 meV atom −1 . We fix this promising architecture (see Methods ) and focus on scaling in the rest of this paper.

Active learning

A core step in our framework for accelerating materials discovery is active learning. In both structural and compositional frameworks, candidate structures filtered using GNoME are evaluated using DFT calculations with standardized settings from the Materials Project. Resulting energies of relaxed structures not only verify the stability of crystal structures but are also incorporated into the iterative active-learning workflow as further training data and structures for candidate generation. Whereas the hit rate for both structural and compositional frameworks start at less than 6% and 3%, respectively, performance improves steadily through six rounds of active learning. Final ensembles of GNoME models improve to a prediction error of 11 meV atom −1 on relaxed structures and hit rates of greater than 80% and 33%, respectively, clearly showing the benefits of scale. An analysis of final GNoME hit rates is provided in Fig. 1d .

Scaling laws and generalization

The test loss performance of GNoME models exhibit improvement as a power law with further data. These results are in line with neural scaling laws in deep learning 28 , 38 and suggest that further discovery efforts could continue to improve generalization. Emphatically, unlike the case of language or vision, in materials science, we can continue to generate data and discover stable crystals, which can be reused to continue scaling up the model. We also demonstrate emergent generalization to out-of-distribution tasks by testing structural models trained on data originating from substitutions on crystals arising from random search 26 in Fig. 1e . These examples are often high-energy local minima and out of distribution compared with data generated by our structural pipeline (which, by virtue of substitutions, contains structures near their minima). Nonetheless, we observe clear improvement with scale. These results indicate that final GNoME models are a substantial step towards providing the community with a universal energy predictor, capable of handling diverse materials structures through deep learning.

Discovered stable crystals

Using the described process of scaling deep learning for materials exploration, we increase the number of known stable crystals by almost an order of magnitude. In particular, GNoME models found 2.2 million crystal structures stable with respect to the Materials Project. Of these, 381,000 entries live on the updated convex hull as newly discovered materials.

Consistent with other literature on structure prediction, the GNoME materials could be bumped off the convex hull by future discoveries, similar to how GNoME displaces at least 5,000 ‘stable’ materials from the Materials Project and the OQMD. See Supplementary Note  1 for discussion on improving structures of already-discovered compositions. Nevertheless, Figs. 1 and 2 provide a summary of the stable materials, with Fig. 1b focusing on the growth over time. We see substantial gains in the number of structures with more than four unique elements in Fig. 2a . This is particularly promising because these materials have proved difficult for previous discovery efforts 27 . Our scaled GNoME models overcome this obstacle and enable efficient discovery in combinatorially large regions.

a , GNoME enables efficient discovery in the combinatorial spaces of 4+ unique elements that can be difficult for human experts. b , Phase-separation energies (energy to the convex hull) for discovered quaternaries showcase similar patterns but larger absolute numbers than previous catalogues. c , Discovered stable crystals correspond to 45,500 novel prototypes as measured by XtalFinder (ref.  39 ). d , Validation by r 2 SCAN shows that 84% of discovered binary and ternary crystals retain negative phase separations with more accurate functionals.

Clustering by means of prototype analysis 39 supports the diversity of discovered crystals with GNoME, leading to more than 45,500 novel prototypes in Fig. 2c (a 5.6 times increase from 8,000 of the Materials Project), which could not have arisen from full substitutions or prototype enumeration. Finally, in Fig. 2b , we compare the phase-separation energy (also referred to as the decomposition enthalpy) of discovered quaternaries with those from the Materials Project to measure the relative distance to the convex hull of all other competing phases. The similarities in distribution suggest that the found materials are meaningfully stable with respect to competing phases and not just ‘filling in the convex hull.’ Further analyses of materials near to (but not on) the updated convex hull is given in Supplementary Note  3 .

Validation through experimental matching and r 2 SCAN

All candidates for GNoME are derived from snapshots of databases made in March 2021, including the Materials Project and the OQMD. Concurrent to our discovery efforts, researchers have continued to experimentally create new crystals, providing a way to validate GNoME findings. Of the experimental structures aggregated in the ICSD, 736 match structures that were independently obtained through GNoME. Six of the experimentally matched structures are presented in Fig. 1c and further details of the experimental matches are provided in Supplementary Note  1 . Similarly, of the 3,182 compositions added to the Materials Project since the snapshot, 2,202 are available in the GNoME database and 91% match on structure. A manual check of ‘newly’ discovered crystals supported the findings, with details in Supplementary Note  4 .

We also validate predictions to ensure that model-based exploration did not overfit simulation parameters. We focus on the choice of functional. Standard projector augmented wave (PAW)-Perdew–Burke–Ernzerhof (PBE) potentials provided a speed–accuracy trade-off suited for large-scale discovery 40 , 41 , but the r 2 SCAN functional provides a more accurate meta-generalized gradient approximation 29 , 42 , 43 . 84% of the discovered binaries and ternary materials also present negative phase-separation energies (as visualized in Fig. 2d , comparable with a 90% ratio in the Materials Project but operating at a larger scale). 86.8% of tested quaternaries also remain stable on the r 2 SCAN convex hull. The discrepancies between PBE and r 2 SCAN energies are further analysed in Supplementary Note  2 .

Composition families of interest

We highlight the benefits of a catalogue of stable materials an order of magnitude larger than previous work. When searching for a material with certain desirable properties, researchers often filter such catalogues, as computational stability is often linked with experimental realizability. We perform similar analyses for three applications. First, layered materials are promising systems for electronics and energy storage 44 . Methods from previous studies 45 suggest that approximately 1,000 layered materials are stable compared with the Materials Project, whereas this number increases to about 52,000 with GNoME-based discoveries. Similarly, following a holistic screening approach with filters such as exclusion of transition metals or by lithium fraction, we find 528 promising Li-ion conductors among GNoME discoveries, a 25 times increase compared with the original study 46 . Finally, Li/Mn transition-metal oxides are a promising family to replace LiCoO 2 in rechargeable batteries 25 and GNoME has discovered an extra 15 candidates stable relative to the Materials Project compared with the original nine.

Scaling up learned interatomic potentials

The process of discovery of stable crystals also provides a data source beyond stable materials. In particular, the ionic relaxations involve computation of first-principles energies and forces for a diverse set of materials structures. This generates a dataset of unprecedented diversity and scale, which we explore to pretrain a general-purpose machine-learning interatomic potential (MLIP) for bulk solids. MLIPs have become a promising tool to accelerate the simulation of materials by learning the energies and forces of reference structures computed at first-principles accuracy 30 , 47 , 48 , 49 . Existing efforts typically train models per material, with data often sampled from ab initio molecular dynamics (AIMD). This markedly limits their general applicability and adoption, requiring expensive data collection and training a new potential from scratch for each system. By making use of the GNoME dataset of first-principles calculations from diverse structural relaxations, we demonstrate that large-scale pretraining of MLIPs enables models that show unprecedented zero-shot accuracy and can be used to discover superionic conductors, without training on any material-specific data.

Zero-shot scaling and generalization

We scale pretraining of a NequIP potential 30 on data sampled from ionic relaxations. Increasing the pretraining dataset, we observe consistent power-law improvements in accuracy (see Fig. 3a,b ). Despite only being trained on ionic relaxations and not on molecular-dynamics data, the pretrained GNoME potential shows remarkable accuracy when evaluated on downstream data sampled from the new distribution of AIMD in a zero-shot manner, that is, in which no training data originate from AIMD simulations (see Fig. 3 ). Notably, this includes unseen compositions, melted structures and structures including vacancies, all of which are not included in our training set (see Supplementary Note  6.4 ). In particular, we find that the scale of the GNoME dataset allows it to outperform existing general-purpose potentials (see Fig. 3d ) and makes the pretrained potential competitive with models trained explicitly on hundreds of samples from the target data distributions (see Supplementary Note  6.4 ). We observe particularly pronounced improvements in the transferability of MLIPs, one of the most pressing shortcomings of MLIPs. To assess the transferability of the potentials, we test their performance under distribution shift: we train two types of NequIP potential on structures sampled from AIMD at T  = 400 K, one in which the network is trained from randomly initialized weights and the other in which we fine-tune from a pretrained GNoME checkpoint. We then measure the performance of both potentials on data sampled from AIMD at T  = 1,000 K (see Fig. 3c ), out of distribution with respective to the 400-K data. The potential pretrained on GNoME data shows systematic and strong improvements in transferability over the potential trained from scratch, even when training is performed on more than 1,000 structures. The zero-shot GNoME potential, not fine-tuned on any data from this composition, outperforms even a state-of-the-art NequIP model trained on hundreds of structures.

a , Classification of whether a material is a superionic conductor as predicted by GNoME-driven simulations in comparison with AIMD, tested on 623 unseen compositions. The classification error improves as a power law with training set size. b , Zero-shot force error as a function of training set size for the unseen material K 24 Li 16 P 24 Sn 8 . c , Robustness under distribution shift, showing the MAE in forces on the example material Ba 8 Li 16 Se 32 Si 8 . A GNoME-pretrained and a randomly initialized potential are trained on data of various sizes sampled at T  = 400 K and evaluated on data sampled at T  = 1,000 K. The zero-shot GNoME potential outperforms state-of-the-art models trained from scratch on hundreds of structures. d , Comparison of zero-shot force errors of three different pretrained, general-purpose potentials for bulk systems on the test set of ref.  56 . Note that the composition Ni is not present in the GNoME pretraining data. RMSE, root-mean-square error.

Screening solid-state ionic conductors

Solid electrolytes are a core component of solid-state batteries, promising higher energy density and safety than liquid electrolytes, but suffer from lower ionic conductivities at present. In the search for novel electrolyte materials, AIMD allows for the prediction of ionic conductivities from first principles. However, owing to the poor scaling of DFT with the number of electrons, routine simulations are limited to hundreds of picoseconds, hundreds of atoms and, most importantly, small compositional search spaces. Here we show that the GNOME potentials show high robustness in this out-of-distribution, zero-shot setting and generalizes to high temperatures, which allows them to serve as a tool for high-throughput discovery of novel solid-state electrolytes. We use GNoME potentials pretrained on datasets of increasing size in molecular-dynamics simulations on 623 never-before-seen compositions. Figure 3a shows the ability of the pretrained GNoME potentials to classify unseen compositions as superionic conductors in comparison with AIMD.

When scaled to the GNoME dataset—much larger than existing approaches—we find that deep learning unlocks previously impossible capabilities for building transferable interatomic potentials for inorganic bulk crystals and allows for high-accuracy, zero-shot prediction of materials properties at scale.

We show that GNNs trained on a large and diverse set of first-principles calculations can enable the efficient discovery of inorganic materials, increasing the number of stable crystals by more than an order of magnitude. Associated datasets empower machine-learned interatomic potentials, giving accurate and robust molecular-dynamics simulations out of the box on unseen bulk materials. Our findings raise interesting questions about the capabilities of deep-learning systems in the natural sciences: the application of machine-learning methods for scientific discovery has traditionally suffered from the fundamental challenge that learning algorithms work under the assumption of identically distributed data at train and test times, but discovery is inherently an out-of-distribution effort. Our results on large-scale learning provide a potential step to move past this dilemma, by demonstrating that GNoME models exhibit emergent out-of-distribution capabilities at scale. This includes discovery in unseen chemical spaces (for example, with more than four different elements), as well as on new downstream tasks (for example, predicting kinetic properties).

GNoME models have already found 2.2 million stable crystals with respect to previous work and enabled previously impossible modelling capabilities for materials scientists. Some open problems remain for the transition of findings in applications, including a greater understanding of phase transitions through competing polymorphs, dynamic stability arising from vibrational profiles and configurational entropies and, ultimately, synthesizability. Nevertheless, we see pretrained, general-purpose GNoME models being used as powerful tools across a diverse range of applications to fundamentally accelerate materials discovery.

Datasets and candidate generation

Snapshots of available datasets.

GNoME discoveries aim to extend the catalogues of known stable crystals. In particular, we build off previous work by the Materials Project 16 , the OQMD 17 , Wang, Botti and Marques (WBM) 27 and the ICSD 15 . For reproducibility, GNoME-based discoveries use snapshots of the two datasets saved at a fixed point in time. We use the data from the Materials Project as of March 2021 and the OQMD as of June 2021. These structures are used as the basis for all discovery including via SAPS, yielding the catalogue of stable crystals as a result of GNoME. Further updates and incorporation of discoveries by these two groups could yield an even greater number of crystal discoveries.

For a revised comparison, another snapshot of the Materials Project, the OQMD and WBM was taken in July 2023. Approximately 216,000 DFT calculations were performed at consistent settings and used to compare the rate of GNoME discoveries versus the rate of discoveries by concurrent research efforts. From 2021 to 2023, the number of stable crystals external to GNoME expanded from 35,000 to 48,000, relatively small in comparison with the 381,000 new stable crystal structures available on the convex hull presented in this paper.

Substitution patterns

Structural substitution patterns are based on data-mined probabilities from ref.  22 . That work introduced a probabilistic model for assessing the likelihood for ionic species substitution within a single crystal structure. In particular, the probability of substitution is calculated as a binary feature model such that \(p(X,{X}^{{\prime} })\approx \frac{\exp {\sum }_{i}{\lambda }_{i}{f}_{i}^{(n)}(X,{X}^{{\prime} })}{Z}\) , in which X and X ′ are n -component vectors of n different ions. The model is simplified so that f i is 0 or 1 if a specific substitution pair occurs and λ i provides a weighting for the likelihood of a given substitution. The resulting probabilities have been helpful, for example, in discovering new quaternary ionic compounds with limited computation budgets.

In our work, we adjust the probabilistic model so as to increase the number of candidates and give priority to discovery. In particular, the conditional probability computation in the original substitution patterns prefers examples that are more likely to be found in the original dataset. For example, any uncommon element is assigned a smaller probability in the original model. To give priority to novel discovery and move further away from the known sets of stable crystals, we modify the implementation so that probabilities are only computed when two compositions differ. This minor modification has substantial benefits across our pipeline, especially when scaling up to six unique elements.

We also introduce changes to the model parameters to promote novel discovery. In the original probabilistic model, positive lambda refers to more likely substitutions, although ‘unseen’ or uncommon substitution resulted in negative lambda values. We increase the number of generations by setting the minimum value of any substitution pair to be 0. We then threshold high-probability substitutions to a value of 0.001, enabling efficient exploration in composition space through branch-and-bound algorithms available from pymatgen. Overall, these settings allow for many one-ion or two-ion substitutions to be considered by the graph networks that otherwise would not have been considered. We find this to be a good intermediate between the original model and using all possible ionic substitutions, in which we encounter combinatorial blow-ups in the number of candidates.

For the main part of this paper, substitutions are only allowed into compositions that do not match any available compositions in the Materials Project or in the OQMD, rather than comparing structures using heuristic structure matchers. This ensures that we introduce novel compositions in the dataset instead of similar structures that may be missed by structure matchers.

To further increase the diversity of structures generations, we introduce a framework that we refer to as symmetry aware partial substitutions (SAPS), which generalizes common substitution frameworks. For a motivating example, consider the cases of (double) perovskites. Ionic substitutions on crystals of composition A 2 B 2 X 6 does not lead to discovering double perovskites A 2 BB′O 6 , although the two only differ by a partial replacement on the B site.

SAPS enable efficient discovery of such structures. Starting with an original composition, we obtain candidate ion replacements using the probabilities as defined in the ‘Substitution patterns’ section. We then obtain Wyckoff positions of the input structures by means of symmetry analysers available through pymatgen. We enable partial replacements from 1 to all atoms of the candidate ion, for which at each level we only consider unique symmetry groupings to control the combinatorial growth. Early experiments limited the partial substitutions to materials that would charge-balance after partial substitutions when considering common oxidation states; however, greater expansion of candidates was achieved by removing such charge-balancing from the later experiments. This partial-substitution framework enables greater use of common crystal structures while allowing for the discovery of new prototypical structures, as discussed in the main part of this paper. Candidates from SAPS are from a different distribution to the candidates from full substitutions, which increases the diversity of our discoveries and our dataset.

To validate the impact of the SAPS, we traced reference structures from substitutions of all 381,000 novel stable structures back to a structure in the Materials Project or the OQMD by means of a topological sort (necessary as discovered materials were recycled for candidate generation). A total of 232,477 out of the 381,000 stable structures can be attributed to a SAPS substitution, suggesting notable benefit from this diverse candidate-generation procedure.

Oxidation-state relaxations

For the compositional pipeline, inputs for evaluation by machine-learning models must be unique stoichiometric ratios between elements. Enumerating the combinatorial number of reduced formulas was found to be too inefficient, but common strategies to reduce such as oxidation-state balancing was also too restrictive, for example, not allowing for the discovery of Li 15 Si 4 . In this paper, we introduce a relaxed constraint on oxidation-state balancing. We start with the common oxidation states from the Semiconducting Materials by Analogy and Chemical Theory (SMACT) 57 , with the inclusion of 0 for metallic forms. We allow for up to two elements to exist between two ordered oxidation states. Although this is a heuristic approach, it substantially improves the flexibility of composition generation around oxidation-state-balanced ratios.

AIRSS structure generation

Random structures are generated through AIRSS when needed for composition models 26 . Random structures are initialized as ‘sensible’ structures (obeying certain symmetry requirements) to a target volume and then relaxed through soft-sphere potentials. A substantial number of initializations and relaxations are needed to discover new materials, as different initial structures lead to different minima on the structure–energy landscape. For this paper, we always generate 100 AIRSS structures for every composition that is otherwise predicted to be within 50 meV of stable through composition-only model prediction.

As we describe in Supplementary Note  5 , not all DFT relaxations converge for the 100 initializations per composition. In fact, for certain compositions, only a few initializations converge. One of the main difficulties arises from not knowing a good initial volume guess for the composition. We try a range of initial volumes ranging from 0.4 to 1.2 times a volume estimated by considering relevant atomic radii, finding that the DFT relaxation fails or does not converge for the whole range for each composition. Prospective analysis was not able to uncover why most AIRSS initializations fail for certain compositions, and future work is needed in this direction.

Model training and evaluation

Graph networks.

For structural models, edges are drawn in the graph when two atoms are closer than an interatomic distance cutoff (4.0 Å for structural models, 5.0 Å for interatomic potentials). Compositional models default to forming edges between all pairs of nodes in the graph. The models update latent node features through stages of message passing, in which neighbour information is collected through normalized sums over edges and representations are updated through shallow MLPs 36 . After several steps of message passing, a linear readout layer is applied to the global state to compute a prediction of the energy.

Training structural and composition models

Following Roost (representation learning from stoichiometry) 58 , we find GNNs to be effective at predicting the formation energy of a composition and structure.

For the structural models, the input is a crystal definition, which encodes the lattice, structure and atom definitions. Each atom is represented as a single node in the graph. Edges are defined when the interatomic distance is less than a user-defined threshold. Nodes are embedded by atom type, edges are embedded on the basis of the interatomic distance. We also include a global feature that is connected in the graph representation to all nodes. At every step of the GNN, neighbouring nodes and edge features are aggregated and used to update the corresponding representations of nodes, edges or globals individually. After 3–6 layers of message passing, an output layer projects the global vector to get an estimate of the energy. All data for training are shifted and scaled to approximately standardize the datasets. This structural model trained on the Materials Project data obtains state-of-the-art results of a mean absolute error of 21 meV atom −1 . Training during the active-learning procedure leads to a model with a final mean absolute error of 11 meV atom −1 . Training for structural models is performed with 1,000 epochs, with a learning rate of 5.55 × 10 −4 and a linear decay learning rate schedule. By default, we train with a batch size of 256 and use swish nonlinearities in the MLP. To embed the edges, we use a Gaussian featurizer. The embedding dimension for all nodes and edges is 256 and, unless otherwise stated, the number of message-passing iterations is 3.

For the compositional models, the input composition to the GNN is encoded as a set of nodes, for which each element type in the composition is represented by a node. The ratio of the specific element is multiplied with the one-hot vector. For example, SiO 2 would be represented with two nodes, in which one node feature is a vector of zeros and a 1/3 on the 14th row to represent silicon and the other node is a vector of zeros with a 2/3 on the 8th row to represent oxygen. Although this simplified GNN architecture is able to achieve state-of-the-art generalization on the Materials Project (MAE of 60 meV atom −1 (ref.  25 )), it does not offer useful predictions for materials discovery, which was also observed by Bartel et al. 25 . One of the issues with compositional models is that they assume that the training label refers to the ground-state phase of a composition, which is not guaranteed for any dataset. Thus, the formation-energy labels in the training and test sets are inherently noisy, and reducing the test error does not necessarily imply that one is learning a better formation-energy predictor. To explore this, we created our own training set of compositional energies, by running AIRSS simulations on novel compositions. As described in Supplementary Note  5 , we find that compositions for which there are only a few completed AIRSS runs tend to have large formation energies, often larger than predicted by the compositional GNN. We find that, if we limit ourselves to compositions for which at least ten AIRSS runs are completed, then the compositional GNN error is reduced to 40 meV atom −1 . We then use the GNN trained on such a dataset (for which labels come from the minimum formation energy phase for compositions with at least ten completed AIRSS runs and ignoring the Materials Project data) and are able to increase the precision of stable prediction to 33%.

Model-based evaluation

Discovering new datasets aided by neural networks requires a careful balance between ensuring that the neural networks trained on the dataset are stable and promoting new discoveries. New structures and prototypes will be inherently out of distribution for models; however, we hope that the models are still capable of extrapolating and yielding reasonable predictions. This is out-of-distribution detection problem is further exacerbated by the implicit domain shift, in which models are trained on relaxed structures but evaluated on substitutions before relaxation. To counteract these effects, we make several adjustments to stabilize test-time predictions.

Test-time augmentations

Augmentations at test time are a common strategy for correcting instabilities in machine-learning predictions. Specific to structural models, we especially consider isotropic scaling of the lattice vectors, which both shrinks and stretches bonds. At 20 values ranging from 80% to 120% of the reference lattice scaling volume, we aggregate by means of minimum reduction. This has the added benefit of potentially correcting for predicting on nonrelaxed structures, as isotropic scaling may yield a more appropriate final structure.

Deep ensembles and uncertainty quantification

Although neural network models offer flexibility that allows them to achieve state-of-the-art performance on a wide range of problems, they may not generalize to data outside the training distribution. Using an ensemble of models is a simple, popular choice for providing predictive uncertainty and improving generalization of machine-learning predictions 33 . This technique simply requires training n models rather than one. The prediction corresponds to the mean over the outputs of all n models; the uncertainty can be measured by the spread of the n outputs. In our application of training machine-learning models for stability prediction, we use n  = 10 graph networks. Moreover, owing to the instability of graph-network predictions, we find the median to be a more reliable predictor of performance and use the interquartile range to bound uncertainty.

Model-based filtration

We use test-time augmentation and deep-ensemble approaches discussed above to filter candidate materials based on energy. Materials are then compared with the available GNoME database to estimate the decomposition energy. Note that the structures provided for model-based filtration are unlikely to be completely related, so a threshold of 50 meV atom −1 was used for active learning to improve the recall of stable crystal discovery.

Clustered-based reduction

For active-learning setups, only the structure predicted to have the minimum energy within a composition is used for DFT verification. However, for an in-depth evaluation of a specific composition family of interest, we design clustering-based reduction strategies. In particular, we take the top 100 structures for any given composition and perform pairwise comparisons with pymatgen’s built-in structure matcher. We cluster the connected components on the graph of pairwise similarities and take the minimum energy structure as the cluster representation. This provides a scalable strategy to discovering polymorphs when applicable.

Active learning was performed in stages of generation and later evaluation of filtered materials through DFT. In the first stage, materials from the snapshots of the Materials Project and the OQMD are used to generate candidates with an initial model trained on the Materials Project data, with a mean absolute error of 21 meV atom −1 in formation energy. Filtration and subsequent evaluation with DFT led to discovery rates between 3% and 10%, depending on the threshold used for discovery. After each round of active learning, new structural GNNs are trained to improve the predictive performance. Furthermore, stable crystal structures are added to the set of materials that can be substituted into, yielding a greater number of candidates to be filtered by the improved models. This procedure of retraining and evaluation was completed six times, yielding the total of 381,000 stable crystal discoveries. Continued exploration with active learning may continue to drive the number of stable crystals higher.

Composition-based hashing

Previous efforts to learn machine-learning models of energies often use a random split over different crystal structures to create the test set on which energy predictions are evaluated. However, as the GNoME dataset contains several crystal structures with the same composition, this metric is less trustworthy over GNoME. Having several structures within the same composition in both the training and the test sets markedly reduces test error, although the test error does not provide a measure of how well the model generalizes to new compositions. In this paper, we use a deterministic hash for the reduced formula of each composition and assign examples to the training (85%) and test (15%) sets. This ensures that there are no overlapping compositions in the training and test sets. We take a standard MD5 hash of the reduced formula, convert the hexadecimal output to an integer and take modulo 100 and threshold at 85.

DFT evaluation

Vasp calculations.

We use the VASP (refs.  34 , 59 ) with the PBE 41 functional and PAW 40 , 60 potentials in all DFT calculations. Our DFT settings are consistent with the Materials Project workflows as encoded in pymatgen 23 and atomate 61 . We use consistent settings with the Materials Project workflow, including the Hubbard U parameter applied to a subset of transition metals in DFT+U, 520 eV plane-wave-basis cutoff, magnetization settings and the choice of PBE pseudopotentials, except for Li, Na, Mg, Ge and Ga. For Li, Na, Mg, Ge and Ga, we use more recent versions of the respective potentials with the same number of valence electrons. For all structures, we use the standard protocol of two-stage relaxation of all geometric degrees of freedom, followed by a final static calculation, along with the custodian package 23 to handle any VASP-related errors that arise and adjust appropriate simulations. For the choice of KPOINTS, we also force gamma-centred kpoint generation for hexagonal cells rather than the more traditional Monkhorst–Pack. We assume ferromagnetic spin initialization with finite magnetic moments, as preliminary attempts to incorporate different spin orderings showed computational costs that were prohibitive to sustain at the scale presented. In AIMD simulations, we turn off spin polarization and use the NVT ensemble with a 2-fs time step.

Bandgap calculations

For validation purposes (such as the filtration of Li-ion conductors), bandgaps are calculated for most of the stable materials discovered. We automate bandgap jobs in our computation pipelines by first copying all outputs from static calculations and using the pymatgen-based MPNonSCFSet in line mode to compute the bandgap and density of states of all materials. A full analysis of patterns in bandgaps of the novel discoveries is a promising avenue for future work.

r 2 SCAN is an accurate and numerically efficient functional that has seen increasing adoption from the community for increasing the fidelity of computational DFT calculations. This functional is provided in the upgraded version of VASP6 and, for all corresponding calculations, we use the settings as detailed by MPScanRelaxSet and MPScanStaticSet in pymatgen. Notably, r 2 SCAN functionals require the use of PBE52 or PBE54 potentials, which can differ slightly from the PBE equivalents used elsewhere in this paper. To speed up computation, we perform three jobs for every SCAN-based computation. First, we precondition by means of the updated PBE54 potentials by running a standard relaxation job under MPRelaxSet settings. This preconditioning step greatly speeds up SCAN computations, which—on average—are five times slower and can otherwise crash on our infrastructure owing to elongated trajectories. Then, we relax with the r 2 SCAN functional, followed by a static computation.

Metrics and analysis methodology

Decomposition energies.

To compute decomposition energies and count the total number of stable crystals relative to previous work 16 , 17 in a consistent fashion, we recalculated energies of all stable materials in the Materials Project and the OQMD with identical, updated DFT settings as enabled by pymatgen. Furthermore, to ensure fair comparison and that our discoveries are not affected by optimization failures in these high-throughput recalculations, we use the minimum energy of the Materials Project calculation and our recalculation when both are available.

Prototype analysis

We validate the novel discoveries using XtalFinder (ref.  39 ), using the compare_structures function available from the command line. This process was parallelized over 96 cores for improved performance. We also note that the symmetry calculations in the built-in library fail on less than ten of the stable materials discovered. We disable these filters but note that the low number of failures suggests minimal impact on the number of stable prototypes.

Families of interest

Layered materials.

To count the number of layered materials, we use the methodology developed in ref.  45 , which is made available through the pymatgen.analysis.dimensionality package with a default tolerance of 0.45 Å.

Li-ion conductors

The estimated number of viable Li-ion conductors reported in the main part of this paper is derived using the methodology in ref.  46 in a high-throughput fashion. This methodology involves applying filters based on bandgaps and stabilities against the cathode Li-metal anode to identify the most viable Li-ion conductors.

Li/Mn transition-metal oxide family

The Li/Mn transition-metal oxide family is discussed in ref.  25 to analyse the capabilities of machine-learning models for use in discovery. In the main text, we compare against the findings in the cited work suggesting limited discovery within this family through previous machine-learning methods.

Definition of experimental match

In the main part of this paper, we refer to experimentally validated crystal structures with the ICSD. More specifically, we queried the ICSD in January 2023 after many of crystal discoveries had been completed. We then extracted relevant journal (year) and chemical (structure) information from the provided files. By rounding to nearest integer formulas, we found 4,235 composition matches with materials discovered by GNoME. Of these, 4,180 are successfully parsed for structure. Then, we turn to the structural information provided by the ICSD. We used the CIF parser module of pymatgen to load the experimental ICSD structures into pymatgen and then compared those to the GNoME dataset using its structure matcher module. For both modules, we tried using the default settings as well as more tolerant settings that improve structure parsing and matching (higher occupancy tolerance in CIF parsing to fix cases with >1.0 total occupancy and allowing supercell and subset comparison in matching). The latter resulted in a slight increase (about 100) in the number of matched structures with respect to the default settings. Given that we are enforcing a strict compositional match, our matching process is still relatively conservative and is likely to yield a lower bound. Overall, we found 736 matches, providing experimental confirmation for the GNoME structures. 184 of these structures correspond to novel discoveries since the start of the project.

Methods for creating figures of GNoME model scaling

Figures 1e and 3a,b show how the generalization abilities of GNoME models scale with training set size. In Fig. 1e , the training sets are sampled uniformly from the materials from the Materials Project and from our structural pipeline, which only includes elemental and partial substitutions into stable materials in the Materials Project and the OQMD. The training labels are the final formation energy at the end of relaxation. The test set is constructed by running AIRSS on 10,000 random compositions filtered by the SMACT. Test labels are the final formation energy at the end of the AIRSS relaxation, for crystals that AIRSS and DFT (both electronically and ionically) converged. Because we apply the same composition-based hash filtering (see ‘Composition-based hashing’ section) on all of our datasets, there is no risk of label leakage between the training set from the structural pipeline and the test set from AIRSS.

In Fig. 3a , we present the classification error for predicting the outcome of DFT-based molecular dynamics using GNN molecular dynamics. ‘GNoME: unique structures’ refers to the first step in the relaxation of crystals in the structural pipeline. We train on the forces on each atom on the first DFT step of relaxation. The different training subsets are created by randomly sampling compositions in the structural pipeline uniformly. ‘GNoME: intermediate structures’ includes all the same compositions as ‘GNoME: unique structures’, but has all steps of DFT relaxation instead of just the first step. The red diamond refers to the same GNN interatomic potential trained on the data from M3GNet, which includes three relaxation steps per composition (first, middle and last), as described in the M3GNet paper 62 .

Coding frameworks

For efforts in machine learning, GNoME models make use of JAX and the capabilities to just-in-time compile programs onto devices such as graphics processing units (GPUs) and tensor processing units (TPUs). Graph networks implementations are based on the framework developed in Jraph, which makes use of a fundamental GraphsTuple object (encoding nodes and edges, along with sender and receiver information for message-passing steps). We also make great of use functionality written in JAX MD for processing crystal structures 63 , as well as TensorFlow for parallelized data input 64 .

Large-scale generation, evaluation and summarization pipelines make use of Apache Beam to distribute processing across a large number of workers and scale to the sizes as described in the main part of this paper (see ‘Overview of generation and filtration’ section). For example, billions of proposal structures, even efficiently encoded, requires terabytes of storage that would otherwise fail on single nodes.

Also, crystal visualizations are created using tooling from VESTA (ref.  65 ).

Pretrained GNoME potential

We train a NequIP potential 30 , implemented in JAX using the e3nn-jax library 66 , with five layers, hidden features of 128 ℓ  = 0 scalars, 64 ℓ  = 1 vectors and 32 ℓ  = 2 tensors (all even irreducible representations only, 128 x 0 e  + 64 x 1 x  + 32 x 2 e ), as well as an edge-irreducible representation of 0 e  + 1 e  + 2 e . We use a radial cutoff of 5 Å and embed interatomic distances r i j in a basis of eight Bessel functions, which is multiplied by the XPLOR cutoff function, as defined in HOOMD-blue (ref.  67 ), using an inner cutoff of 4.5 Å. We use a radial MLP R ( r ) with two hidden layers with 64 neurons and a SiLU nonlinearity. We also use SiLU for the gated, equivariant nonlinearities 68 . We embed the chemical species using a 94-element one-hot encoding and use a self-connection, as proposed in ref.  30 . For internal normalization, we divide by 26 after each convolution. Models are trained with the Adam optimizer using a learning rate of 2 × 10 −3 and a batch size of 32. Given that high-energy structures in the beginning of the trajectory are expected to be more diverse than later, low-energy structures, which are similar to one another and often come with small forces, each batch is made up of 16 structures sampled from the full set of all frames across all relaxations and 16 structures sampled from only the first step of the relaxation only. We found this oversampling of first-step structures to substantially improve performance on downstream tasks. The learning rate was decreased to a new value of 2 × 10 −4 after approximately 23 million steps, to 5 × 10 −5 after a further approximately 11 million steps and then trained for a final 2.43 million steps. Training was performed on four TPU v3 chips.

We train on formation energies instead of total energies. Formation energies and forces are not normalized for training but instead we predict the energy as a sum over scaled and shifted atomic energies, such that \(\widehat{E}={\sum }_{i\in {N}_{{\rm{atoms}}}}\left({\widehat{{\epsilon }}}_{i}\sigma +\mu \right)\) , in which \({\widehat{{\epsilon }}}_{i}\) is the final, scalar node feature on atom i and σ and μ are the standard deviation and mean of the per-atom energy computed over a single pass of the full dataset. The network was trained on a joint loss function consisting of a weighted sum of a Huber loss on energies and forces:

in which N a and N b denote the number of atoms in a structure and the number of samples in a batch, respectively, \({\widehat{E}}_{{\rm{b}}}\) and E b are the predicted and true energy for a given sample in a batch, respectively, and F a , α is the true force component on atom a , for which α   ∈  { x ,  y ,  z } is the spatial component. \({{\mathcal{L}}}_{{\rm{Huber}}}(\delta ,\widehat{a},a)\) denotes a Huber loss on quantity a , for which we use δ E = δ F = 0.01. The pretrained potential has 16.24 million parameters. Inference on an A100 GPU on a 50-atom system takes approximately 14 ms, enabling a throughput of approximately 12 ns day −1 at a 2-fs time step, making inference times highly competitive with other implementations of GNN interatomic potentials. Exploring new approaches with even further improved computational efficiency is the focus of future work.

Training on M3GNet data

To allow a fair comparison with the smaller M3GNet dataset used in ref.  62 , a NequIP model was trained on the M3GNet dataset. We chose the hyperparameters in a way that balances accuracy and computational efficiency, resulting in a potential with efficient inference. We train in two setups, one splitting the training and testing sets based on unique materials and the other over all structures. In both cases, we found the NequIP potential to perform better than the M3GNet models trained with energies and forces (M3GNet-EF) reported in ref.  62 . Given this improved performance, to enable a fair comparison of datasets and dataset sizes, we use the NequIP model trained on the structure-split M3GNet data in the scaling tests (the pretrained M3GNet model is used for zero-shot comparisons). We expect our scaling and zero-shot results to be applicable to a wide variety of modern deep-learning interatomic potentials.

The structural model used for downstream evaluation was trained using the Adam optimizer with a learning rate of 2 × 10 −3 and a batch size of 16 for a total of 801 epochs. The learning rate was decreased to 2 × 10 −4 after 601 epochs, after which we trained for another 200 epochs. We use the same joint loss function as in the GNoME pretraining, again with λ E  = 1.0, λ F  = 0.05 and δ E  =  δ F  = 0.01. The network hyperparameters are identical to the NequIP model used in GNoME pretraining. To enable a comparison with ref.  62 , we also subtract a linear compositional fit based on the training energies from the reference energies before training. Training was performed on a set of four V100 GPUs.

AIMD conductivity experiments

Following ref.  69 , we classify a material as having superionic behaviour if the conductivity σ at the temperature of 1,000 K, as measured by AIMD, satisfies σ 1,000K  > 101.18 mScm −1 . Refer to the original paper for applicable calculations. See  Supplementary Information for further details.

Robustness experiments

For the materials selected for testing the robustness of our models, As 24 Ca 24 Li 24 , Ba 8 Li 16 Se 32 Si 8 , K 24 Li 16 P 24 Sn 8 and Li 32 S 24 Si 4 , a series of models is trained on increasing training set sizes sampled from the T  = 400 K AIMD trajectory. We then evaluate these models on AIMD data sampled at both T  = 400 K (to measure the effect of fine-tuning on data from the target distribution) and T  = 1,000 K (to measure the robustness of the learned potentials). We trained two types of model: (1) a NequIP model from scratch and (2) a fine-tuned model that was pretrained on the GNoME dataset, starting from the checkpoint before the learning rate was reduced the first time. The network architecture is identical to that used in pretraining. Because the AIMD data contain fewer high-force/high-energy configurations, we use a L2 loss in the joint loss function instead of a Huber loss, again with λ E  = 1.0 and λ F  = 0.05. For all training set sizes and all materials, we scan learning rates 1 × 10 −2 and 2 × 10 −3 and batch sizes 1 and 16. Models are trained for a maximum of 1,000 epochs. The learning rate is reduced by a factor of 0.8 if the test error on a hold-out set did not improve for 50 epochs. We choose the best of these hyperparameters based on the performance of the final checkpoint on the 400-K test set. The 400-K test set is created using the final part of the AIMD trajectory. The training sets are created by sampling varying training set sizes from the initial part of the AIMD trajectory. The out-of-distribution robustness test is generated from the AIMD trajectory at 1,000 K. Training is performed on a single V100 GPU.

Molecular dynamics simulations

The materials for AIMD simulation are chosen on the basis of the following criteria: we select all materials in the GNoME database that are stable, contain one of the conducting species under consideration (Li, Mg, Ca, K, Na) and have a computationally predicted band gap >1 eV. The last criterion is chosen to not include materials with notable electronic conductivity, a desirable criterion in the search for electrolytes. Materials are run in their pristine structure, that is, without vacancies or stuffing. The AIMD simulations were performed using the VASP. The temperature is initialized at T  = 300 K, ramped up over a time span of 5 ps to the target temperature, using velocity rescaling. This is followed by a 45-ps simulation equilibration using a Nosé–Hoover thermostat in the NVT ensemble. Simulations are performed at a 2-fs time step.

Machine-learning-driven molecular dynamics simulations using JAX MD 63 are run on a subset of materials for which AIMD data were available and for which the composition was in the test set of the pretraining data (that is, previously unseen compositions), containing Li, Na, K, Mg and Ca as potentially conducting species. This results in 623 materials for which GNoME-driven molecular dynamics simulations are run. Simulations are performed at T  =1,000 K using a Nosé–-Hoover thermostat, a temperature equilibration constant of 40 time steps, a 2-fs time step and a total simulation length of 50 ps. Molecular dynamics simulations are performed on a single P100 GPU.

For analysis of both the AIMD and the machine learning molecular dynamics simulation, the first 10 ps of the simulation are discarded for equilibration. From the final 40 ps, we compute the diffusivity using the DiffusionAnalyzer class of pymatgen with the default smoothed=max setting 23 , 70 , 71 .

Data availability

Crystal structures corresponding to stable discoveries discussed throughout the paper will be made available at https://github.com/google-deepmind/materials_discovery . In particular, we provide results for all stable structures, as well as any material that has been recomputed from previous datasets to ensure consistent settings. Associated data from the r 2 SCAN functional will be provided, expectantly serving as a foundation for analysing discrepancies between functional choices. Data will also be available via the Materials Project at https://materialsproject.org/gnome with permanent link: https://doi.org/10.17188/2009989 .

Code availability

Software to analyse stable crystals and associated phase diagrams, as well as the software implementation of the static GNN and the interatomic potentials, will be made available at https://github.com/google-deepmind/materials_discovery .

Green, M. A., Ho-Baillie, A. & Snaith, H. J. The emergence of perovskite solar cells. Nat. Photon.   8 , 506–514 (2014).

Article   ADS   CAS   Google Scholar  

Mizushima, K., Jones, P., Wiseman, P. & Goodenough, J. B. Li x CoO 2 (0< x <-1): a new cathode material for batteries of high energy density. Mater. Res. Bull. 15 , 783–789 (1980).

Article   CAS   Google Scholar  

Bednorz, J. G. & Müller, K. A. Possible high T c superconductivity in the Ba–La–Cu–O system. Z. Phys. B Condens. Matter 64 , 189–193 (1986).

Ceder, G. et al. Identification of cathode materials for lithium batteries guided by first-principles calculations. Nature 392 , 694–696 (1998).

Tabor, D. P. et al. Accelerating the discovery of materials for clean energy in the era of smart automation. Nat. Rev. Mater. 3 , 5–20 (2018).

Liu, C. et al. Two-dimensional materials for next-generation computing technologies. Nat. Nanotechnol. 15 , 545–557 (2020).

Article   ADS   CAS   PubMed   Google Scholar  

Nørskov, J. K., Bligaard, T., Rossmeisl, J. & Christensen, C. H. Towards the computational design of solid catalysts. Nat. Chem. 1 , 37–46 (2009).

Article   PubMed   Google Scholar  

Greeley, J., Jaramillo, T. F., Bonde, J., Chorkendorff, I. & Nørskov, J. K. Computational high-throughput screening of electrocatalytic materials for hydrogen evolution. Nat. Mater. 5 , 909–913 (2006).

Gómez-Bombarelli, R. et al. Design of efficient molecular organic light-emitting diodes by a high-throughput virtual screening and experimental approach. Nat. Mater. 15 , 1120–1127 (2016).

Article   ADS   PubMed   Google Scholar  

de Leon, N. P. et al. Materials challenges and opportunities for quantum computing hardware. Science 372 , eabb2823 (2021).

Wedig, A. et al. Nanoscale cation motion in TaO x , HfO x and TiO x memristive systems. Nat. Nanotechnol. 11 , 67–74 (2016).

Brown, T. et al. Language models are few-shot learners. Adv. Neural Inf. Process. Syst. 33 , 1877–1901 (2020).

Google Scholar  

Dosovitskiy, A. et al. An image is worth 16x16 words: Transformers for image recognition at scale. In International Conference on Learning Representations (ICLR, 2021); https://openreview.net/forum?id=YicbFdNTTy

Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596 , 583–589 (2021).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Hellenbrandt, M. The Inorganic Crystal Structure Database (ICSD)—present and future. Crystallogr. Rev. 10 , 17–22 (2004).

Jain, A. et al. Commentary: The Materials Project: a materials genome approach to accelerating materials innovation. APL Mater. 1 , 011002 (2013).

Article   ADS   Google Scholar  

Saal, J. E., Kirklin, S., Aykol, M., Meredig, B. & Wolverton, C. Materials design and discovery with high-throughput density functional theory: the Open Quantum Materials Database (OQMD). JOM 65 , 1501–1509 (2013).

Belsky, A., Hellenbrandt, M., Karen, V. L. & Luksch, P. New developments in the Inorganic Crystal Structure Database (ICSD): accessibility in support of materials research and design. Acta Crystallogr. B Struct. Sci. 58 , 364–369 (2002).

Aykol, M., Montoya, J. H. & Hummelshøj, J. Rational solid-state synthesis routes for inorganic materials. J. Am. Chem. Soc. 143 , 9244–9259 (2021).

Article   CAS   PubMed   Google Scholar  

Curtarolo, S. et al. AFLOWLIB.ORG: a distributed materials properties repository from high-throughput ab initio calculations. Comput. Mater. Sci. 58 , 227–235 (2012).

Draxl, C. & Scheffler, M. The NOMAD laboratory: from data sharing to artificial intelligence. J. Phys. Mater. 2 , 036001 (2019).

Hautier, G., Fischer, C., Ehrlacher, V., Jain, A. & Ceder, G. Data mined ionic substitutions for the discovery of new compounds. Inorg. Chem. 50 , 656–663 (2011).

Ong, S. P. et al. Python Materials Genomics (pymatgen): a robust, open-source Python library for materials analysis. Comput. Mater. Sci. 68 , 314–319 (2013).

Aykol, M. et al. Network analysis of synthesizable materials discovery. Nat. Commun. 10 , 2018 (2019).

Article   ADS   PubMed   PubMed Central   Google Scholar  

Bartel, C. J. et al. A critical examination of compound stability predictions from machine-learned formation energies. npj Comput. Mater. 6 , 97 (2020).

Pickard, C. J. & Needs, R. Ab initio random structure searching. J. Phys. Condens. Matter 23 , 053201 (2011).

Wang, H.-C., Botti, S. & Marques, M. A. Predicting stable crystalline compounds using chemical similarity. npj Comput. Mater. 7 , 12 (2021).

Hestness, J. et al. Deep learning scaling is predictable, empirically. Preprint at https://arxiv.org/abs/1712.00409 (2017).

Furness, J. W., Kaplan, A. D., Ning, J., Perdew, J. P. & Sun, J. Accurate and numerically efficient r 2 SCAN meta-generalized gradient approximation. J. Phys. Chem. Lett. 11 , 8208–8215 (2020).

Batzner, S. et al. E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials. Nat. Commun. 13 , 2453 (2022).

Thomas, N. et al. Tensor field networks: rotation- and translation-equivariant neural networks for 3D point clouds. Preprint at https://arxiv.org/abs/1802.08219 (2018).

Togo, A. & Tanaka, I. Spglib: a software library for crystal symmetry search. Preprint at https://arxiv.org/abs/1808.01590 (2018).

Behler, J. Constructing high-dimensional neural network potentials: a tutorial review. Int. J. Quantum Chem. 115 , 1032–1050 (2015).

Kresse, G. & Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54 , 11169 (1996).

Battaglia, P. W. et al. Relational inductive biases, deep learning, and graph networks. Preprint at https://arxiv.org/abs/1806.01261 (2018).

Gilmer, J., Schoenholz, S. S., Riley, P. F., Vinyals, O. & Dahl, G. E. Neural message passing for quantum chemistry. Proc. Mach. Learn. Res. 70 , 1263–1272 (2017).

Chen, C., Ye, W., Zuo, Y., Zheng, C. & Ong, S. P. Graph networks as a universal machine learning framework for molecules and crystals. Chem. Mater. 31 , 3564–3572 (2019).

Kaplan, J. et al. Scaling laws for neural language models. Preprint at https://arxiv.org/abs/2001.08361 (2020).

Hicks, D. et al. AFLOW-XtalFinder: a reliable choice to identify crystalline prototypes. npj Comput. Mater. 7 , 30 (2021).

Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50 , 17953 (1994).

Perdew, J. P., Ernzerhof, M. & Burke, K. Rationale for mixing exact exchange with density functional approximations. J. Chem. Phys. 105 , 9982–9985 (1996).

Kitchaev, D. A. et al. Energetics of MnO 2 polymorphs in density functional theory. Phys. Rev. B 93 , 045132 (2016).

Kingsbury, R. et al. Performance comparison of r 2 SCAN and SCAN metaGGA density functionals for solid materials via an automated, high-throughput computational workflow. Phys. Rev. Mater. 6 , 013801 (2022).

Bassman Oftelie, L. et al. Active learning for accelerated design of layered materials. npj Comput. Mater. 4 , 74 (2018).

Cheon, G. et al. Data mining for new two- and one-dimensional weakly bonded solids and lattice-commensurate heterostructures. Nano Lett. 17 , 1915–1923 (2017).

Sendek, A. D. et al. Holistic computational structure screening of more than 12000 candidates for solid lithium-ion conductor materials. Energy Environ. Sci. 10 , 306–320 (2017).

Behler, J. & Parrinello, M. Generalized neural-network representation of high-dimensional potential-energy surfaces. Phys. Rev. Lett. 98 , 146401 (2007).

Bartók, A. P., Payne, M. C., Kondor, R. & Csányi, G. Gaussian approximation potentials: the accuracy of quantum mechanics, without the electrons. Phys. Rev. Lett. 104 , 136403 (2010).

Lot, R., Pellegrini, F., Shaidu, Y. & Küçükbenli, E. PANNA: properties from artificial neural network architectures. Comput. Phys. Commun. 256 , 107402 (2020).

Article   MathSciNet   CAS   Google Scholar  

Zhou, Y., Qiu, Y., Mishra, V. & Mar, A. Lost horses on the frontier: K 2 BiCl 5 and K 3 Bi 2 Br 9 . J. Solid State Chem. 304 , 122621 (2021).

Abudurusuli, A. et al. Li 4 MgGe 2 S 7 : the first alkali and alkaline-earth diamond-like infrared nonlinear optical material with exceptional large band gap. Angew. Chem. Int. Ed. 60 , 24131–24136 (2021).

Ruan, B.-B., Yang, Q.-S., Zhou, M.-H., Chen, G.-F. & Ren, Z.-A. Superconductivity in a new T 2 -phase Mo 5 GeB 2 . J. Alloys Compd. 868 , 159230 (2021).

Guo, Z. et al. Local distortions and metal–semiconductor–metal transition in quasi-one-dimensional nanowire compounds AV 3 Q 3 O δ (A = K, Rb, Cs and Q = Se, Te). Chem. Mater. 33 , 2611–2623 (2021).

Deng, A. et al. Novel narrow-band blue light-emitting phosphor of Eu 2+ -activated silicate used for WLEDs. Dalton Trans. 50 , 16377–16385 (2021).

Zhak, O., Köhler, J., Karychort, O. & Babizhetskyy, V. New ternary phosphides RE 5 Pd 9 P 7 ( RE =Tm, Lu): synthesis, crystal and electronic structure. Z. Anorg. Allg. Chem. 648 , e202200024 (2022).

Zuo, Y. et al. Performance and cost assessment of machine learning interatomic potentials. J. Phys. Chem. A 124 , 731–745 (2020).

Davies, D. W. et al. SMACT: semiconducting materials by analogy and chemical theory. J. Open Source Softw. 4 , 1361 (2019).

Goodall, R. E. & Lee, A. A. Predicting materials properties without crystal structure: deep representation learning from stoichiometry. Nat. Commun. 11 , 6280 (2020).

Kresse, G. & Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 6 , 15–50 (1996).

Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B 59 , 1758 (1999).

Mathew, K. et al. atomate: a high-level interface to generate, execute, and analyze computational materials science workflows. Comput. Mater. Sci. 139 , 140–152 (2017).

Chen, C. & Ong, S. P. A universal graph deep learning interatomic potential for the periodic table. Nat. Comput. Sci. 2 , 718–728 (2022).

Article   Google Scholar  

Schoenholz, S. & Cubuk, E. D. JAX MD: a framework for differentiable physics. Adv. Neural Inf. Process. Syst. 33 , 11428–11441 (2020).

MATH   Google Scholar  

Abadi, M. et al. TensorFlow: large-scale machine learning on heterogeneous systems. https://www.tensorflow.org/ (2015).

Momma, K. & Izumi, F. VESTA 3 for three-dimensional visualization of crystal, volumetric and morphology data. J. Applied Crystallogr. 44 , 1272–1276 (2011).

Geiger, M. & Smidt, T. e3nn: Euclidean neural networks. Preprint at https://arxiv.org/abs/2207.09453 (2022).

Anderson, J. A., Glaser, J. & Glotzer, S. C. HOOMD-blue: a Python package for high-performance molecular dynamics and hard particle Monte Carlo simulations. Comput. Mater. Sci. 173 , 109363 (2020).

Hendrycks, D. & Gimpel, K. Gaussian Error Linear Units (GELUs). Preprint at https://arxiv.org/abs/1606.08415 (2016).

Jun, K. et al. Lithium superionic conductors with corner-sharing frameworks. Nat. Mater. 21 , 924–931 (2022).

Ong, S. P. et al. Phase stability, electrochemical stability and ionic conductivity of the Li 10±1 MP 2 X 1 2 (M = Ge, Si, Sn, Al or P, and X = O, S or Se) family of superionic conductors. Energy Environ. Sci. 6 , 148–156 (2013).

Mo, Y., Ong, S. P. & Ceder, G. First principles study of the Li 1 0GeP 2 S 1 2 lithium super ionic conductor material. Chem. Mater. 24 , 15–17 (2012).

Download references

Acknowledgements

We would like to acknowledge D. Eck, J. Sohl-Dickstein, J. Dean, J. Barral, J. Shlens, P. Kohli and Z. Ghahramani for sponsoring the project; L. Dorfman for product management support; A. Pierson for programme management support; O. Loum for help with computing resources; L. Metz for help with infrastructure; E. Ocampo for help with early work on the AIRSS pipeline; A. Sendek, B. Yildiz, C. Chen, C. Bartel, G. Ceder, J. Sun, J. P. Holt, K. Persson, L. Yang, M. Horton and M. Brenner for insightful discussions; and the Google DeepMind team for continuing support.

Author information

These authors contributed equally: Amil Merchant, Simon Batzner, Samuel S. Schoenholz, Ekin Dogus Cubuk

Authors and Affiliations

Google DeepMind, Mountain View, CA, USA

Amil Merchant, Simon Batzner, Samuel S. Schoenholz, Muratahan Aykol & Ekin Dogus Cubuk

Google Research, Mountain View, CA, USA

Gowoon Cheon

You can also search for this author in PubMed   Google Scholar

Contributions

A.M. led the code development, experiments and analysis in most parts of the project, including the proposal of the data flywheel through active learning, candidate generation (for example, invention of SAPS), large-scale training and evaluation workflows, DFT calculations, convex-hull analysis and materials screening. S.B. led the code development, training and experiments of the force fields and the zero-shot evaluations, fine-tuning, robustness and the GNN molecular dynamics experiments, and contributed to overall code development, as well as training infrastructure. S.S.S. led the scaling of GNN training and JAX MD infrastructure and contributed to force-field experiments. M.A. contributed to data analyses, validation and benchmarking efforts, ran experiments and provided guidance. G.C. contributed to analysis, zero-shot evaluations and provided guidance. E.D.C. conceived and led the direction of the project, wrote software for data generation, model implementations and training, and led the scaling experiments. All authors contributed to discussion and writing.

Corresponding authors

Correspondence to Amil Merchant or Ekin Dogus Cubuk .

Ethics declarations

Competing interests.

Google LLC owns intellectual property rights related to this work, including, potentially, patent rights.

Peer review

Peer review information.

Nature thanks the anonymous reviewers for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information.

The supplementary information contains six sections, providing further context to the computational experiments performed.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and Permissions

About this article

Cite this article.

Merchant, A., Batzner, S., Schoenholz, S.S. et al. Scaling deep learning for materials discovery. Nature (2023). https://doi.org/10.1038/s41586-023-06735-9

Download citation

Received : 08 May 2023

Accepted : 10 October 2023

Published : 29 November 2023

DOI : https://doi.org/10.1038/s41586-023-06735-9

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Google ai and robots join forces to build new materials.

• Mark Peplow

Nature (2023)

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

• Original Paper
• Published: 01 December 2023

Adverse Childhood Experiences and Muscle Dysmorphia Symptomatology: Findings from a Sample of Canadian Adolescents and Young Adults

• Kyle T. Ganson   ORCID: orcid.org/0000-0003-3889-3716 1 ,
• Nelson Pang 1 ,
• Alexander Testa 2 ,
• Dylan B. Jackson 3 &
• Jason M. Nagata 4  

Clinical Social Work Journal ( 2023 ) Cite this article

128 Altmetric

Metrics details

Adverse childhood experiences (ACEs) are relatively common among the general population and have been shown to be associated with eating disorders and body dysmorphic disorder. It remains relatively unknown whether ACEs are associated with muscle dysmorphia. The aim of this study was to investigate the association between ACEs and muscle dysmorphia symptomatology among a sample of Canadian adolescents and young adults. A community sample of 912 adolescents and young adults ages 16–30 years across Canada participated in this study. Participants completed a 15-item measure of ACEs (categorized to 0, 1, 2, 3, 4, and 5 or more) and the Muscle Dysmorphic Disorder Inventory. Multiple linear regression analyses were utilized to determine the association between the number of ACEs experienced and muscle dysmorphia symptomatology. Participants who experienced five or more ACEs, compared to those who had experienced no ACEs, had more symptoms of muscle dysmorphia, as well as more symptoms related to Appearance Intolerance and Functional Impairment. There was no association between ACEs and Drive for Size symptoms. Participants who experienced five or more ACEs (16.1%), compared to 10.6% who experienced no ACEs, were at clinical risk for muscle dysmorphia (p = .018). Experiencing ACEs, particularly five or more, was significantly associated with muscle dysmorphia symptomatology, expanding prior research on eating disorders and body dysmorphic disorder. Social workers should consider screening for symptoms of muscle dysmorphia among adolescents and young adults who experience ACEs.

This is a preview of subscription content, access via your institution .

Access options

Buy single article.

Instant access to the full article PDF.

Price excludes VAT (USA) Tax calculation will be finalised during checkout.

Rent this article via DeepDyve.

Data Availability

Data may be made available upon reasonable request.

American Psychological Association. (2017). Clinical practice guideline for the treatment of posttraumatic stress disorder (PTSD) in Adults . American Psychological Association.

Google Scholar  

American Psychiatric Association. (2022). Diagnostic and statistical manual of mental disorders (5th ed.). American Psychiatric Association.

Book   Google Scholar  

Afifi, T. O., MacMillan, H. L., Boyle, M., Taillieu, T., Cheung, K., & Sareen, J. (2014). Child abuse and mental disorders in Canada. Canadian Medical Association Journal, 186 (9), E324–E332. https://doi.org/10.1503/cmaj.131792

Article   PubMed   PubMed Central   Google Scholar  

Allison, P. D. (2002). Missing data . SAGE Publications Inc.

Bödicker, C., Reinckens, J., Höfler, M., & Hoyer, J. (2022). Is childhood maltreatment associated with body image disturbances in adulthood? A systematic review and meta-analysis. Journal of Child and Adolescent Trauma, 15 (3), 523–538. https://doi.org/10.1007/s40653-021-00379-5

Article   PubMed   Google Scholar  

Brooke, L., & Mussap, A. J. (2012). Brief report: Maltreatment in childhood and body concerns in adulthood. Journal of Health Psychology, 18 (5), 620–626. https://doi.org/10.1177/1359105312454036

Brown, S. M., Bender, K., Orsi, R., McCrae, J. S., Phillips, J. D., & Rienks, S. (2019). Adverse childhood experiences and their relationship to complex health profiles among child welfare–involved children: A classification and regression tree analysis. Health Services Research, 54 (4), 902–911. https://doi.org/10.1111/1475-6773.13166

Cafri, G., Olivardia, R., & Thompson, J. K. (2008). Symptom characteristics and psychiatric comorbidity among males with muscle dysmorphia. Comprehensive Psychiatry, 49 (4), 374–379. https://doi.org/10.1016/j.comppsych.2008.01.003

Centers for Disease Control and Prevention (2019). Preventing adverse childhood experiences (ACEs): Leveraging the best available evidence . pp. 1–40

Chu, J., Raney, J. H., Ganson, K. T., Wu, K., Rupanagunta, A., Testa, A., Jackson, D. B., Murray, S. B., & Nagata, J. M. (2022). Adverse childhood experiences and binge—eating disorder in early adolescents. Journal of Eating Disorders . https://doi.org/10.1186/s40337-022-00682-y

Cohen, J. A., & Mannarino, A. P. (2015). Trauma-focused cognitive behavior therapy for traumatized children and families. Child and Adolescent Psychiatric Clinics of North America, 24 (3), 557–570. https://doi.org/10.1016/j.chc.2015.02.005

Compte, E. J., Cattle, C. J., Lavender, J. M., Murray, S. B., Brown, T. A., Capriotti, M. R., Flentje, A., Lubensky, M. E., Obedin-Maliver, J., Lunn, M. R., & Nagata, J. M. (2021). Psychometric evaluation of the muscle dysmorphic disorder inventory (MDDI) among cisgender gay men and cisgender lesbian women. Body Image, 38 , 241–250. https://doi.org/10.1016/j.bodyim.2021.04.008

Didie, E. R., Tortolani, C. C., Pope, C. G., Menard, W., Fay, C., & Phillips, K. A. (2006). Childhood abuse and neglect in body dysmorphic disorder. Child Abuse and Neglect, 30 (10), 1105–1115. https://doi.org/10.1016/j.chiabu.2006.03.007

England-Mason, G., Casey, R., Ferro, M., MacMillan, H. L., Tonmyr, L., & Gonzalez, A. (2018). Child maltreatment and adult multimorbidity: Results from the Canadian community health survey. Canadian Journal of Public Health, 109 (4), 561–572. https://doi.org/10.17269/s41997-018-0069-y

Fabris, M. A., Badenes-Ribera, L., & Longobardi, C. (2021). Bullying victimization and muscle dysmorphic disorder in Italian adolescents: The mediating role of attachment to peers. Children and Youth Services Review, 120 (November 2020), 105720. https://doi.org/10.1016/j.childyouth.2020.105720

Article   Google Scholar  

Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., Koss, M. P., & Marks, J. S. (2019). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: The adverse childhood experiences (ACE) study. American Journal of Preventive Medicine, 56 (6), 774–786. https://doi.org/10.1016/j.amepre.2019.04.001

Florida Department of Children and Families (2022). 2022 Florida youth substance abuse survey . https://www.myflfamilies.com/services/samh/2022-florida-youth-substance-abuse-survey

Frederick, D. A., Shapiro, L. M., Williams, T. R., Seoane, C. M., McIntosh, R. T., & Fischer, E. W. (2017). Precarious manhood and muscularity: Effects of threatening men’s masculinity on reported strength and muscle dissatisfaction. Body Image, 22 , 156–165. https://doi.org/10.1016/j.bodyim.2017.07.002

Frederick, J., Spratt, T., & Devaney, J. (2021). Adverse childhood experiences and social work: Relationship-based practice responses. British Journal of Social Work, 51 (8), 3018–3034. https://doi.org/10.1093/bjsw/bcaa155

Ganson, K. T. (2022). Canadian study of adolescent health behaviors: Technical report . https://socialwork.utoronto.ca/wp-content/uploads/2022/11/Canadian-Study-of-Adolescent-Health-Behaviors-Technical-Report.pdf

Ganson, K. T., Cunningham, M. L., Pila, E., Rodgers, R. F., Murray, S. B., & Nagata, J. M. (2022). Bulking and cutting among a national sample of Canadian adolescents and young adults. Eating and Weight Disorders Studies on Anorexia Bulimia and Obesity . https://doi.org/10.1007/s40519-022-01470-y

Ganson, K. T., Cunningham, M. L., Pila, E., Rodgers, R. F., Murray, S. B., & Nagata, J. M. (2022a). Characterizing cheat meals among a national sample of Canadian adolescents and young adults. Journal of Eating Disorders . https://doi.org/10.1186/s40337-022-00642-6

Ganson, K. T., Hallward, L., Cunningham, M. L., Murray, S. B., & Nagata, J. M. (2022c). Anabolic-androgenic steroid use: Patterns of use among a national sample of Canadian adolescents and young adults. Performance Enhancement and Health . https://doi.org/10.1016/j.peh.2022.100241

Ganson, K. T., Hallward, L., Cunningham, M. L., Murray, S. B., & Nagata, J. M. (2023). Use of legal appearance- and performance-enhancing drugs and substances: findings from the Canadian study of adolescent health behaviors. Substance Use & Misuse, 58 (2), 289–297. https://doi.org/10.1080/10826084.2022.2161318

Ganson, K. T., Hallward, L., Cunningham, M. L., Rodgers, R. F., & Nagata, J. M. (2023). Muscle dysmorphia symptomatology among a national sample of Canadian adolescents and young adults. Body Image, 44 (8), 178–186. https://doi.org/10.1016/j.bodyim.2023.01.001

Ganson, K. T., Hallward, L., Rodgers, R. F., Testa, A., Jackson, D. B., & Nagata, J. M. (2023). Associations between violent victimization and symptoms of muscle dysmorphia: Findings from the Canadian study of adolescent health behaviors. Body Image, 46 , 294–299. https://doi.org/10.1016/j.bodyim.2023.06.014

Ganson, K. T., Murray, S. B., Mitchison, D., Hawkins, M. A. W., Layman, H., Tabler, J., & Nagata, J. M. (2021). Associations between adverse childhood experiences and performance-enhancing substance use among young adults. Substance Use and Misuse, 56 (6), 854–860. https://doi.org/10.1080/10826084.2021.1899230

Ganson, K. T., Nagata, J. M., Lavender, J. M., Rodgers, R. F., Cunningham, M., Murray, S. B., & Hammond, D. (2021). Prevalence and correlates of weight gain attempts across five countries. International Journal of Eating Disorders . https://doi.org/10.1002/eat.23595

Ganson, K. T., Rodgers, R. F., Austin, S. B., Murray, S. B., & Nagata, J. M. (2022). Prevalence and sociodemographic characteristics of moderate and high engagement in muscle-building exercise among adolescents. Body Image, 42 , 263–267. https://doi.org/10.1016/j.bodyim.2022.07.004

Ganson, K. T., Testa, A., Rodgers, R. F., Jackson, D. B., & Nagata, J. M. (2023d). Relationships between violent sexual victimization and muscle-building exercise among adolescents from the 2019 youth risk behavior survey. Journal of School Health . https://doi.org/10.1111/josh.13395

Garcia, A. R., Gupta, M., Greeson, J. K. P., Thompson, A., & DeNard, C. (2017). Adverse childhood experiences among youth reported to child welfare: Results from the national survey of child & adolescent wellbeing. Child Abuse and Neglect, 70 (January), 292–302. https://doi.org/10.1016/j.chiabu.2017.06.019

Gruber, A. J., & Pope, H. G. (1999). Compulsive weight lifting and anabolic drug abuse among women rape victims. Comprehensive Psychiatry, 40 (4), 273–277. https://doi.org/10.1016/S0010-440X(99)90127-X

Grunewald, W., & Blashill, A. J. (2021). Muscle dysmorphia . Springer.

Grunewald, W., Troop-Gordon, W., & Smith, A. R. (2022). Relationships between eating disorder symptoms, muscle dysmorphia symptoms, and suicidal ideation: A random intercepts cross-lagged panel approach. International Journal of Eating Disorders, 55 (12), 1733–1743. https://doi.org/10.1002/eat.23819

Guillaume, S., Jaussent, I., Maimoun, L., Ryst, A., Seneque, M., Villain, L., Hamroun, D., Lefebvre, P., Renard, E., & Courtet, P. (2016). Associations between adverse childhood experiences and clinical characteristics of eating disorders. Scientific Reports, 6 , 1–7. https://doi.org/10.1038/srep35761

Harrison, A., de la Fernández, L., Enander, J., Radua, J., & Mataix-Cols, D. (2016). Cognitive-behavioral therapy for body dysmorphic disorder: A systematic review and meta-analysis of randomized controlled trials. Clinical Psychology Review, 48 , 43–51. https://doi.org/10.1016/j.cpr.2016.05.007

Hildebrandt, T., Langenbucher, J., & Schlundt, D. G. (2004). Muscularity concerns among men: Development of attitudinal and perceptual measures. Body Image, 1 (2), 169–181. https://doi.org/10.1016/j.bodyim.2004.01.001

Hughes, K., Bellis, M. A., Hardcastle, K. A., Sethi, D., Butchart, A., Mikton, C., Jones, L., & Dunne, M. P. (2017). The effect of multiple adverse childhood experiences on health: A systematic review and meta-analysis. The Lancet Public Health, 2 (8), e356–e366. https://doi.org/10.1016/S2468-2667(17)30118-4

Hunger, J. M., Smith, J. P., & Tomiyama, A. J. (2020). An evidence-based rationale for adopting weight-inclusive health policy. Social Issues and Policy Review, 14 (1), 73–107. https://doi.org/10.1111/sipr.12062

Ip, E. J., Barnett, M. J., Tenerowicz, M. J., & Perry, P. J. (2011). The anabolic 500 survey: Characteristics of male users versus nonusers of anabolic-androgenic steroids for strength training. Pharmacotherapy, 31 (8), 757–766. https://doi.org/10.1592/phco.31.8.757

Joshi, D., Raina, P., Tonmyr, L., MacMillan, H. L., & Gonzalez, A. (2021). Prevalence of adverse childhood experiences among individuals aged 45 to 85 years: A cross-sectional analysis of the Canadian longitudinal study on aging. CMAJ Open, 9 (1), E158–E166. https://doi.org/10.9778/cmajo.20200064

Kessler, R. C., Amminger, P., Aguilar-Gaxiola, S., Alonso, J., Lee, S., & Ustan, B. (2007). Age of onset of mental disorders: A review of recent literature. Current Opinion in Psychiatry, 20 (4), 359–364.

Knight, C. (2015). Trauma-informed social work practice: Practice considerations and challenges. Clinical Social Work Journal, 43 (1), 25–37. https://doi.org/10.1007/s10615-014-0481-6

LaBrenz, C. A., O’Gara, J. L., Panisch, L. S., Baiden, P., & Larkin, H. (2020). Adverse childhood experiences and mental and physical health disparities: The moderating effect of race and implications for social work. Social Work in Health Care, 59 (8), 588–614. https://doi.org/10.1080/00981389.2020.1823547

Lacey, R. E., & Minnis, H. (2020). Practitioner review: Twenty years of research with adverse childhood experience scores – advantages, disadvantages and applications to practice. Journal of Child Psychology and Psychiatry, 61 (2), 116–130. https://doi.org/10.1111/jcpp.13135

Larkin, H., Felitti, V. J., & Anda, R. F. (2014). Social work and adverse childhood experiences research: Implications for practice and health policy. Social Work in Public Health, 29 (1), 1–16. https://doi.org/10.1080/19371918.2011.619433

Levenson, J. (2017). Trauma-informed social work practice. Social Work, 62 (2), 105–113. https://doi.org/10.1093/sw/swx001

Longobardi, C., Badenes-Ribera, L., & Fabris, M. A. (2022). Adverse childhood experiences and body dysmorphic symptoms: A meta-analysis. Body Image, 40 , 267–284. https://doi.org/10.1016/j.bodyim.2022.01.003

Longobardi, C., Prino, L. E., Fabris, M. A., & Settanni, M. (2017). Muscle dysmorphia and psychopathology: Findings from an Italian sample of male bodybuilders. Psychiatry Research, 256 (June), 231–236. https://doi.org/10.1016/j.psychres.2017.06.065

Malcolm, A., Pikoos, T. D., Grace, S. A., Castle, D. J., & Rossell, S. L. (2021). Childhood maltreatment and trauma is common and severe in body dysmorphic disorder. Comprehensive Psychiatry, 109 , 152256. https://doi.org/10.1016/j.comppsych.2021.152256

McKay, M. T., Kilmartin, L., Meagher, A., Cannon, M., Healy, C., & Clarke, M. C. (2022). A revised and extended systematic review and meta-analysis of the relationship between childhood adversity and adult psychiatric disorder. Journal of Psychiatric Research, 156 (October), 268–283. https://doi.org/10.1016/j.jpsychires.2022.10.015

Mitchison, D., Mond, J., Griffiths, S., Hay, P., Nagata, J. M., Bussey, K., Trompeter, N., Lonergan, A., & Murray, S. B. (2021). Prevalence of muscle dysmorphia in adolescents: Findings from the Everybody study. Psychological Medicine . https://doi.org/10.1017/S0033291720005206

Murray, S. B., Nagata, J. M., Griffiths, S., Calzo, J. P., Brown, T. A., Mitchison, D., Blashill, A. J., & Mond, J. M. (2017). The enigma of male eating disorders: A critical review and synthesis. Clinical Psychology Review, 57 (August), 1–11. https://doi.org/10.1016/j.cpr.2017.08.001

Murray, S. B., Rieger, E., Touyz, S. W., & García, Y. D. L. G. (2010). Muscle dysmorphia and the DSM-V conundrum: Where does it belong? A review paper. International Journal of Eating Disorders, 43 (6), 483–491. https://doi.org/10.1002/eat.20828

Nagata, J. M., Bibbins-Domingo, K., Garber, A. K., Griffiths, S., Vittinghoff, E., & Murray, S. B. (2019). Boys, bulk, and body ideals: Sex differences in weight-gain attempts among adolescents in the United States. Journal of Adolescent Health, 64 (4), 450–453. https://doi.org/10.1016/j.jadohealth.2018.09.002

Nagata, J. M., Compte, E. J., Cattle, C. J., Lavender, J. M., Brown, T. A., Murray, S. B., Flentje, A., Capriotti, M. R., Lubensky, M. E., Obedin-Maliver, J., & Lunn, M. R. (2021). Community norms of the muscle dysmorphic disorder inventory (MDDI) among cisgender sexual minority men and women. BMC Psychiatry, 21 (1), 1–9. https://doi.org/10.1186/s12888-021-03302-2

Nagata, J. M., Compte, E. J., McGuire, F. H., Lavender, J. M., Brown, T. A., Murray, S. B., Flentje, A., Capriotti, M. R., Lubensky, M. E., Obedin-Maliver, J., & Lunn, M. R. (2021). Community norms of the muscle dysmorphic disorder inventory (MDDI) among gender minority populations. Journal of Eating Disorders, 9 (1), 1–10. https://doi.org/10.1186/s40337-021-00442-4

Nagata, J. M., Compte, E. J., McGuire, F. H., Lavender, J. M., Murray, S. B., Brown, T. A., Capriotti, M. R., Flentje, A., Lubensky, M. E., Obedin-Maliver, J., & Lunn, M. R. (2022). Psychometric validation of the muscle dysmorphic disorder inventory (MDDI) among U.S. transgender men. Body Image, 42 , 43–49. https://doi.org/10.1016/j.bodyim.2022.05.001

Nagata, J. M., Ganson, K. T., Griffiths, S., Mitchison, D., Garber, A. K., Vittinghoff, E., Bibbins-Domingo, K., & Murray, S. B. (2020). Prevalence and correlates of muscle-enhancing behaviors among adolescents and young adults in the United States. International Journal of Adolescent Medicine and Health . https://doi.org/10.1515/ijamh-2020-0001

Nagata, J. M., Ganson, K. T., & Murray, S. B. (2020). Eating disorders in adolescent boys and young men: An update. Current Opinion in Pediatrics, 32 (4), 476–481. https://doi.org/10.1097/MOP.0000000000000911

Olivardia, R., Pope, J., & Hudson, J. I. (2000). Muscle dysmorphia in male weightlifters: A case-control study. American Journal of Psychiatry, 157 (8), 1291–1296. https://doi.org/10.1176/appi.ajp.157.8.1291

Oral, R., Ramirez, M., Coohey, C., Nakada, S., Walz, A., Kuntz, A., Benoit, J., & Peek-Asa, C. (2016). Adverse childhood experiences and trauma informed care: The future of health care. Pediatric Research, 79 (1–2), 227–233. https://doi.org/10.1038/pr.2015.197

Ortiz, S. N., Forrest, L. N., & Smith, A. R. (2021). Correlates of suicidal thoughts and attempts in males engaging in muscle dysmorphia or eating disorder symptoms. Journal of Clinical Psychology, 77 (4), 1106–1115. https://doi.org/10.1002/jclp.23102

Ozan Tingaz, E. (2020). Association between muscle dysmorphia and childhood abuse and neglect in male recreational bodybuilders. Journal on Educational Psychology, 13 (4), 19.

Parent, M. C. (2013). Handling item-level missing data: Simpler is just as good. The Counseling Psychologist, 41 (4), 568–600. https://doi.org/10.1177/0011000012445176

Petruccelli, K., Davis, J., & Berman, T. (2019). Adverse childhood experiences and associated health outcomes: A systematic review and meta-analysis. Child Abuse and Neglect, 97 (February), 104127. https://doi.org/10.1016/j.chiabu.2019.104127

Pope, C. G., Pope, H. G., Menard, W., Fay, C., Olivardia, R., & Phillips, K. A. (2005). Clinical features of muscle dysmorphia among males with body dysmorphic disorder. Body Image, 2 (4), 395–400. https://doi.org/10.1038/jid.2014.371

Pope, H. G., Gruber, A. J., Choi, P., Olivardia, R., & Phillips, K. A. (1997). Muscle dysmorphia: An underrecognized form of body dysmorphic disorder. Psychosomatics, 38 (6), 548–557. https://doi.org/10.1016/S0033-3182(97)71400-2

Pope, H. G., Khalsa, J. H., & Bhasin, S. (2017). Body image disorders and abuse of anabolic-androgenic steroids among men. JAMA Journal of the American Medical Association, 317 (1), 23–24. https://doi.org/10.1001/jama.2016.17441

Rienecke, R. D., Johnson, C., Le Grange, D., Manwaring, J., Mehler, P. S., Duffy, A., McClanahan, S., & Blalock, D. V. (2022). Adverse childhood experiences among adults with eating disorders: Comparison to a nationally representative sample and identification of trauma. Journal of Eating Disorders, 10 (1), 1–10. https://doi.org/10.1186/s40337-022-00594-x

Siegel, B. S., Dobbins, M. I., Earls, M. F., Garner, A. S., Pascoe, J., Wood, D. L., High, P. C., Donoghue, E., Fussell, J. J., Gleason, M. M., Jaudes, P. K., Jones, V. F., Rubin, D. M., Schulte, E. E., Macias, M. M., Bridgemohan, C., Goldson, E., McGuinn, L. J., Weitzman, C., & Wegner, L. M. (2012). Early childhood adversity, toxic stress, and the role of the pediatrician: Translating developmental science into lifelong health. Pediatrics . https://doi.org/10.1542/peds.2011-2662

Solmi, M., Radua, J., Olivola, M., Croce, E., Soardo, L., de Salazar, G., Il Shin, J., Kirkbride, J. B., Jones, P., Kim, J. H., Kim, J. Y., Carvalho, A. F., Seeman, M. V., Correll, C. U., & Fusar-Poli, P. (2021). Age at onset of mental disorders worldwide: Large-scale meta-analysis of 192 epidemiological studies. Molecular Psychiatry . https://doi.org/10.1038/s41380-021-01161-7

Spratt, T., Devaney, J., & Frederick, J. (2019). Adverse childhood experiences: Beyond signs of safety; reimagining the organisation and practice of social work with children and families. The British Journal of Social Work, 49 (8), 2042–2058. https://doi.org/10.1093/bjsw/bcz023

Tiggemann, M., & Zaccardo, M. (2018). Strong is the new skinny: A content analysis of #fitspiration images on Instagram. Journal of Health Psychology, 23 (8), 1003–1011. https://doi.org/10.1177/1359105316639436

Tod, D., Edwards, C., & Cranswick, I. (2016). Muscle dysmorphia: Current insights. Psychology Research and Behavior Management, 9 , 179–188. https://doi.org/10.2147/PRBM.S97404

Vandello, J. A., Bosson, J. K., Cohen, D., Burnaford, R. M., & Weaver, J. R. (2008). Precarious manhood. Journal of Personality and Social Psychology, 95 (6), 1325–1339. https://doi.org/10.1037/a0012453

Varangis, E., Folberth, W., Hildebrandt, T., Langenbucher, J. (2012). Confirmatory factor analysis for the muscle dysmorphic disorder inventory among male appearance and performance enhancing drug users. Austin: International Conference on Eating Disorders

Vartanian, L. R., Hayward, L. E., Smyth, J. M., Paxton, S. J., & Touyz, S. W. (2018). Risk and resiliency factors related to body dissatisfaction and disordered eating: The identity disruption model. International Journal of Eating Disorders, 51 (4), 322–330. https://doi.org/10.1002/eat.22835

Wilhelm, S., Phillips, K. A., Greenberg, J. L., O’Keefe, S. M., Hoeppner, S. S., Keshaviah, A., Sarvode-Mothi, S., & Schoenfeld, D. A. (2019). Efficacy and posttreatment effects of therapist-delivered cognitive behavioral therapy vs supportive psychotherapy for adults with body dysmorphic disorder: A randomized clinical trial. JAMA Psychiatry, 02114 , 363–373. https://doi.org/10.1001/jamapsychiatry.2018.4156

Xu, Y., Pace, S., Kim, J., Iachini, A., King, L. B., Harrison, T., DeHart, D., Levkoff, S. E., Browne, T. A., Lewis, A. A., Kunz, G. M., Reitmeier, M., Utter, R. K., & Simone, M. (2022). Threats to online surveys: Recognizing, detecting, and preventing survey bots. Social Work Research . https://doi.org/10.1093/swr/svac023

Zeeck, A., Welter, V., Alatas, H., Hildebrandt, T., Lahmann, C., & Hartmann, A. (2018). Muscle dysmorphic disorder inventory (MDDI): Validation of a German version with a focus on gender. PLoS ONE, 13 (11), 1–13. https://doi.org/10.1371/journal.pone.0207535

Download references

Acknowledgements

This study was funded by the Connaught New Researcher Award (#512586) at the University of Toronto (KTG). JMN was funded by the National Institutes of Health (K08HL159350 and R01MH135492).

Author information

Authors and affiliations.

Factor-Inwentash Faculty of Social Work, University of Toronto, 246 Bloor Street W, Toronto, ON, M5S 1V4, Canada

Kyle T. Ganson & Nelson Pang

Department of Management, Policy and Community Health, University of Texas Health Science Center at Houston, Houston, TX, USA

Alexander Testa

Department of Population, Family, and Reproductive Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA

Dylan B. Jackson

Department of Pediatrics, University of California, San Francisco, San Francisco, CA, USA

Jason M. Nagata

You can also search for this author in PubMed   Google Scholar

Contributions

KTG: Funding acquisition, Project administration, Conceptualization, Investigation, Methodology, Formal analysis, Writing—Original draft, Writing—Review & editing; NP: Writing—Original draft, Writing—Review & editing; AT: Conceptualization, Writing—Review & editing; DBJ: Conceptualization, Writing—Review & editing; JMN: Conceptualization, Writing—Review & editing.

Corresponding author

Correspondence to Kyle T. Ganson .

Ethics declarations

Conflict of interest.

All authors report no conflict of interest.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary material 1 (DOCX 2887.9 kb)

Rights and permissions.

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and Permissions

About this article

Cite this article.

Ganson, K.T., Pang, N., Testa, A. et al. Adverse Childhood Experiences and Muscle Dysmorphia Symptomatology: Findings from a Sample of Canadian Adolescents and Young Adults. Clin Soc Work J (2023). https://doi.org/10.1007/s10615-023-00908-9

Download citation

Accepted : 10 November 2023

Published : 01 December 2023

DOI : https://doi.org/10.1007/s10615-023-00908-9

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Adverse childhood experiences
• Muscle dysmorphia
• Adolescents
• Young adults

Advertisement

• Find a journal
• Publish with us

Duke Neurology Research Round Up, December 2023

What do a speech prosthetic that translates brain signals into speech, retinal scans that detect cognitive impairment, and a promising new form of genetic therapy for Parkinson’s and some forms of dementia have in common? They’re all examples of the 21 peer-reviewed journal articles authored members of the Duke Neurology Department published this November.  

Other highlights include insights into how the brain codes different types of memory, recommendations for ending daylight savings time to improve health, and discussions the potential to use modulation to improve recovery from stroke. Read the paragraphs below to find brief summaries of all of these articles and more, as well as links to the original research.

Epilepsy, Clinical Neurophysiology, and Sleep

• Senior authors Rahul Gaini, MD, and Elijah Lackey, MD, as well as Duke medical student Julia Denniss and Ashley Lengel, MS, PA-C, wrote a new case report exploring the clinical presentation and genetic findings of a 44-year-old male with a history of pediatric epilepsy. The case report also discusses a novel gene variant shared by the patient and his daughter and how that variant influenced his neurologic symptoms. Read it here in Cureus. 
• Gregory Cogan, PhD, led a collaborative team of Duke neuroscientists, neurosurgeons, and engineers that developed a speech prosthetic that can translate a person’s brain signals into what they’re trying to say. The new technology might one day help people unable to talk due to neurological disorders regain the ability to communicate through a brain-computer interface. Read that article in Nature Communications.
• When daylight savings time ended this year on November 5, the American Academy of Sleep Medicine issued a position statement arguing that permanent standard time is the optimal choice for both health and safety. Andrew Spector, MD, contributed to that statement. Read it here.
• Shruti Agashe, MD,  was the first author on a paper that provided a comprehensive review of intracranial hemorrhage in sEEG regardless of clinical symptoms. She also identified electrographic correlates of hemorrhage to help in early detection. Read that article in the Journal of Clinical Neurophysiology.

Memory Disorders

• A novel artificial intelligence model utilizes retinal scans from widely available imaging technology to distinguish individuals with mild cognitive impairment (MCI) from those with normal cognition, thanks to a collaborative study with the Duke Department of Ophthalmology. Senior author Sharon Fekrat, MD FASRS, along with Andrew Liu, MD, MS, Kim Johnson, MD, and other Duke colleagues in the iMIND Study Group, developed a multimodal convolutional neural network (CNN) that was trained, validated, and tested using these multimodal retinal images along with quantitative data to identify those with MCI. Read that article in Ophthalmology Science.
• Senior author Sharon Fekrat, MD, and Kim Johnson, MD, were part of a team that examined differences in retinal and choroidal microvasculature and structure in individuals with dementia with Lewy bodies (DLB) and individuals with normal cognition. Fekrat’s team used optical coherence tomography (OCT) and OCT angiography to analyze images from patients with DLB and patients with normal cognition, finding that patients with DLB had an increased peripapillary CPD, decreased peripapillary CFI, and attenuated GC-IPL thickness. Read the full article in the Journal of VitreoRetinal Diseases.
• Effective, accurate clinical trials are needed to develop new therapies for Alzheimer’s disease, but the slow, varied progression of this condition makes these trials difficult. Kathleen Welsh-Bohmer, PhD, was the first author of a new article that addresses these challenges as well as new directions and actionable steps to improve future trial designs. Read that article in Alzheimer’s & Dementia.

Neurocritical Care

• A new article in Critical Care Medicine examines associations of early sedation patterns, as well as the association of dexmedetomidine exposure, with clinical and functional outcomes among mechanically ventilated patients with moderate-severe traumatic brain injury (msTBI). Katharine Colton, MD, Daniel Laskowitz, MD, MHS,and M. Luke James, MD, contributed to that article, which found variation in early sedation choice among mechanically ventilated patients and a lack of improved 6-month functional outcomes among patients with early dexmedetomidine exposure. Read that article here.

Neuro-Oncology

• Tumors of the central nervous system are the most common pediatric cancer in the United States, but most physicians caring for these patients are not formally certified in this subspecialty. Katherine Peters, MD, PhD, was part of a team that examined physician, patient, and caregiver support for formal certification in neuro-oncology. Read what they found in Neuro-Oncology Advances.
• Suma Shah, MD, was part of a randomized controlled trial investigating the effects of cryocompression therapy on chemotherapy-induced peripheral neuropathy. Shah and colleagues enrolled 91 gynecologic cancer patients planned for five to six cycles of neurotoxic chemotherapy and assigned them to receive cryocompression on their dominant hand or foot, and no cryocompression on the opposite limb. The therapy decreased the odds of patients’ sensory neuropathy by 46% at their final visit. Read the full article in Obstetrics & Gynecology.

Neuromuscular Disease

• Senior author Rick Bedlack, MD, PhD, Xiaoyan Li, MD, and Tasnim Mushannen, MD, contributed to the 72nd entry in the ALSUntangled series, which reviews alternative and off-label treatments for people living with amyotrophic lateral sclerosis (ALS). In this entry, Bedlack and colleagues examine plausible mechanisms, and existing data for the use of insulin in slowing ALS progression. Read their article in Amyotrophic Lateral Sclerosis and Frontotemporal Degeneration.
• A new case review in the Journal of Neuromuscular Disease provides insights into the course of double-seronegative myasthenia gravis (DSNMG) during and after pregnancy. Senior author Donald Sanders, MD, Janice Massey, MD, Vern Juel, MD, and colleagues performed a  retrospective cohort study of women with DSNMG seen in the Duke Myasthenia Gravis Clinic over the past 20 years. Their analysis found women with DSNMG had increased MG symptoms during pregnancy and within 6 months postpartum, similar to seropositive MG. Read that article here.

Stroke and Vascular Neurology

• Shreyansh Shah, MD, was part of a team that sheds light on current patterns associated with the use of antithrombotic and statin therapies after intracerebral hemorrhage (ICH). The team’s analysis of nearly 500,000 patients with ICH found about 10% were prescribed antiplatelet medications, 35% were prescription statins, and 5% were prescribed anticoagulation therapy at discharge. Read the full article in Stroke.
• Each year, more than 600,000 people suffer serious, often permanent disabilities due to stroke. Adjunctive neuromodulation is an emerging therapy that offers the potential to help reduce these disabilities and improve stroke survivors’ quality of life. A new review article summarizes and discusses various neuromodulation techniques, including existing clinical evidence and their future potential. Senior author Wuwei “Wayne” Feng, MD, MS, and Shashank Shekhar, MD, contributed to that article, which appears in Current Neurology and Neuroscience Reports.
• Senior author Wuwei “Wayne” Feng, MD, MS, Sara Hassani, MD, Dylan Ryan, MD and colleagues performed a systematic review and meta analysis examining the use of propranolol or beta blockers for cerebral cavernous malformation. Read that study in Translational Stroke Research.
• Scott Le, DO, and Rahul Gaini, MD, wrote a case report detailing an elderly woman who was found to have caseous calcification of the papillary muscle (CCPM) after a possible central retinal artery occlusion or eye stroke. Their article provides more details on her diagnosis and treatment as well as how CCPM may act as a risk factor for cardioembolic stroke. Read the full article  in BMJ Case Reports.

Translational Brain Sciences

• New research offers a promising new therapeutic avenue for treating Parkinson’s disease and dementia with Lewy bodies. Senior author Ornit Chiba-Falek, PhD, Duke Neurobiology’s Boris Kantor, PhD, and Zhigou Sun, PhD, describe a novel neuronal-type specific epigenome therapy that reduces overexpression of SNCA, the gene whose overexpression leads to the development of these conditions. The team also demonstrated in vitro proof-of-concept using human based disease models. Future research will examine the potential for this therapy in preclinical animal models and, eventually, clinical trials. Read the article in Molecular Therapy Nucleic Acids.
• Tatiana Segura, PhD, led the development of two novel biomaterial formulations of granular hydrogels for tissue regeneration after stroke: highly porous microgels and microgels loaded with heparin-norbornene nanoparticles with covalently bound SDF-1α. These hydrogels may improve our ability to deliver stem cells, growth factors, or other therapies to improve tissue repair and outcomes for stroke survivors. Read more about them in Advanced Healthcare Materials.
• Extended reality (XR) technologies have shown promise to help people living with autism spectrum disorders, but there gaps in our understanding of the neurobiology of autism, particularly relating to sex-based differences, have limited this potential. Estate Sokhadze, PhD, was the senior author of a new review article synthesizing the current research on brain activity patterns in autism spectrum disorder, emphasizing the implications for XR interventions and neurofeedback therapy. Read that article here.
• Simon Davis, PhD, was the senior author of a new study that provides fresh insights into how our brains remember. Davis and colleagues used functional MRI to observe 19 study participants as they tried to memorize images of real-world objects like pizza, birds, and clothing. They later tested the participants on their memory of both the concept of the objects and the specific details.They found that certain brain regions exhibited a heightened sensitivity to visual characteristics, while others were more attuned to semantic, or meaning-related, information. Read the full article in the Journal of Clinical Neuroscience.

Other Topics

• Sneha Mantri, MD, MS, contributed to a new article examining the psychological impact of the COVID-19 epidemic on the prevalence of moral injury, burnout, depression, and anxiety in the healthcare workforce. Mantri and colleagues administered a series of surveys to 17,000 employees of a large academic medical between December 2021 and February 2022. Across all roles, the prevalence of moral injury, burnout, depression, and anxiety were 41%, 35%-61%, 25%, and 25%, respectively. Read the full article in the Journal of Nervous and Mental Disease.

Three Ways to Tell If Research Is Bunk

Here are some rules for deciding whether a new social-science finding is really useful to you.

Want to stay current with Arthur’s writing? Sign up to get an email every time a new column comes out.

W e live in interesting times for the social sciences. For the past several decades, disciplines such as social psychology and behavioral economics seemed to unlock many of the mysteries of human life, and academics and laypeople alike couldn’t get enough of what they revealed. Journalists, too, lapped up insights into how, say, otherwise decent people given arbitrary power over others can become brutal and even sadistic—as the famous Stanford prison experiment purported to find when it asked volunteers to simulate the roles of prisoners and prison guards. And everyone delighted in the cleverly eye-catching ways researchers designed such studies.

Lately, however, the social-science world has become mired in controversy. Researchers themselves have started to note that many famous experiments have been debunked —such as, indeed, the Stanford prison experiment—or simply can’t be replicated. Scholars writing in the journal Science in 2015 reproduced 100 experiments published in three highly influential psychology journals and found that just 36 percent yielded results consistent with the original findings.

Worse still, multiple allegations of unethical data practices have emerged, some proven. The misconduct has included so-called p-hacking —processing data in search of spurious correlations and publishing them as valid results—and even outright fraud , in which researchers have altered their data to fit a preconceived result.

Read: An unsettling hint at how much fraud could exist in science

A natural conclusion for many outside the profession is grave skepticism about the field: People are bound to wonder if research based on behavioral experiments simply can’t be trusted. I get that, but I still reject the notion that the whole enterprise has been discredited. As an academic social scientist—and, more to the point, an arbiter of such research for readers of The Atlantic —I’d like to offer a look behind the curtain, to show how research works, why it goes wrong, how to read it, what to trust, and what to disregard.

M isbegotten studies are not limited to the social sciences. The natural sciences suffer from very similar difficulties. John Ioannidis, a medical researcher at Stanford University, attributes this problem to a series of study flaws: experiments that are too small to trust; cases in which a result seems valid because of mathematical sleight of hand but is effectively meaningless; poor experimental designs; and—the most entrenched—academic bias caused by career incentives to find particular results.

These incentives are brutal. In some fields, a professor needs to have published dozens of research articles to have a prayer of getting tenure (without which they are usually shown the door). Imagine studying for a decade through college and graduate school, then beavering away for perhaps another 10 years on research while managing a hefty teaching load, only to be in the position of betting your career on the up-or-out decision of a university committee. As one researcher wrote on Political Science Rumors, an anonymous message board where academics in this field can openly converse, “The tenure track is a dehumanizing process, because it treats us as apprentices until we’re at least well into our 30s and often significantly past 40.” Although my own quest for tenure was years ago, I well recall the 70-hour weeks in a windowless office and intense stress that it involved.

At top universities, even for published research to be properly executed is not good enough. The topic also has to be clever, and ideally the findings should be surprising. Empirical evidence that dog bites man doesn’t get much credit; an ingenious experiment that uncovers evidence suggesting that, actually, men bite dogs 73.4 percent of the time—now we’re talking.

Not surprisingly, perhaps, this impulse to center on the catchily counterintuitive leads to problems. A 2019 article titled “Cheat or Perish?” in the journal Research Policy looked at the incentives in academia to cut corners or straight-up cheat. Its authors noted that although carrots can encourage productive work that is high quality, even in outwardly dispassionate research, sticks can deter mistakes or misbehavior. But the scholars also wrote that errors and misconduct in research can be hard to detect, which effectively reduces the ability to investigate them. This means that the carrots heavily outweigh the sticks, predictably affecting the integrity of the work.

Tenure alone does not necessarily eliminate the problematic incentives. Senior researchers easily fall prey to hubris and fail to seek advice and constructive criticism. Research has also shown that people with more expertise in a field tend to react more poorly than those with less experience by becoming even more pedantic when they receive “disconfirming feedback”—in other words, they don’t want to hear that they’re wrong. And the stakes rise in an era when behavioral scientists can become superstars outside academia, earning speaking engagements and book deals for their thrillingly heterodox findings.

I do not believe that most experimental research in the social sciences is false, and I’ve been blessed to work with colleagues and collaborators throughout my career who are uncompromisingly scrupulous. Yet I am very cautious about the research that I cite in this column, in addition to our careful routine fact-checking process, precisely because of the problems described above. In preparing each week, I usually get the lay of a research landscape and then closely read 10 or 12 relevant academic articles. When I’m deciding what to use to make my arguments, I first look at the quality of the research design, using my academic training. Then I follow three basic rules.

1. If it seems too good to be true, it probably is.

Over the past few years, three social scientists—Uri Simonsohn, Leif Nelson, and Joseph Simmons—have become famous for their sleuthing to uncover false or faked research results. To make the point that many apparently “legitimate” findings are untrustworthy, they tortured one particular data set until it showed the obviously impossible result that listening to the Beatles song “When I’m Sixty-Four” could literally make you younger.

So if a behavioral result is extremely unusual, I’m suspicious. If it is implausible or runs contrary to common sense, I steer clear of the finding entirely, because the risk that it is false is too great. I like to subject behavioral science to what I call the “grandparent test”: Imagine describing the result to your worldly-wise older relative, and getting their response. (“Hey, Grandma, I found a cool new study showing that infidelity leads to happier marriages. What do you think?”)

2. Let ideas age a bit.

I tend to trust a sweet spot for how recent a particular research finding is. A study published more than 20 years ago is usually too old to reflect current social circumstances. But if a finding is too new, it may have so far escaped sufficient scrutiny—and been neither replicated nor shredded by other scholars. Occasionally, a brand-new paper strikes me as so well executed and sensible that it is worth citing to make a point, and I use it, but I am generally more comfortable with new- ish studies that are part of a broader pattern of results in an area I am studying. I keep a file (my “wine cellar”) of very recent studies that I trust but that I want to age a bit before using for a column.

3. Useful beats clever.

The perverse incentive is not limited to the academy alone. A lot of science journalism values novelty over utility, reporting on studies that turn out to be more likely to fail when someone tries to replicate them. As well as leading to confusion, this misunderstands the point of behavioral science, which is to provide not edutainment but insights that can improve well-being.

I rarely write a column because I find an interesting study . Instead, I come across an interesting topic or idea and write about that. Then I go looking for answers based on a variety of research and evidence. That gives me a bias—for useful studies over clever ones.

B eyond checking the methods, data, and design of studies, I feel that these three rules work pretty well in a world of imperfect research. In fact, they go beyond how I do my work; they actually help guide how I live.

In life, we’re constantly beset by fads and hacks—new ways to act and think and be, shortcuts to the things we want. Whether in politics, love, faith, or fitness, the equivalent of some hot new study with counterintuitive findings is always demanding that we throw out the old ways and accept the latest wisdom.

Read: Online bettors can sniff out weak psychology studies

I believe in personal progress. That is why I write this column, and I like to think I’ve made some in my life. But I also know that our novelty- and technology-obsessed culture is brimming with bad ideas and misinformation—the equivalent of p-hacked results, false findings, and outright fraud for personal gain.

So, in life as in work, when I see a bandwagon going by, I always ask, Does this seem too good to be true? I let the cultural moment or social trend age for a while. And then I consider whether it is useful, as opposed to simply novel. Maybe this pause makes me a bit of a square—but it rarely fails me.

• Open access
• Published: 05 December 2023

Treatment of brainstem and fourth ventricle lesions by the full neuroendoscopic telovelar approach

• Long Zhou 1   na1 ,
• Hangyu Wei 1   na1 ,
• Zhiyang Li 1   na1 ,
• Huikai Zhang 1 ,
• Ping Song 1 ,
• Li Cheng 2 ,
• Wenju Wang 1 ,
• Pan Lei 1 ,
• Qianxue Chen 1 ,
• Zaiming Liu 1 ,
• Daofa Sun 3 &
• Qiang Cai 1  

European Journal of Medical Research volume  28 , Article number:  564 ( 2023 ) Cite this article

Metrics details

To explore the surgical techniques, advantages, and disadvantages of neuroendoscopic telovelar approach in the treatment of brainstem and fourth ventricle lesions.

The clinical data of 5 patients treated by neuroendoscopic telovelar approach from March 2020 to March 2022 were analyzed retrospectively.

Among the 5 patients, there were 3 cavernous hemangiomas in pontine arm and 2 tumors in brainstem and fourth ventricle. All patients could successfully complete the operation, and 4 patients recovered well, other 1 patient discharged automatically for serious complications of other systems after the operation.

The telovelar approach has gained popularity as a safe and effective strategy for lesions in fourth ventricular and brainstem. However, without removing the posterior arch of the atlas, it is difficult to enter the upper part of the fourth ventricle under a microscope. Transcranial neuroendoscopy can effectively compensate for the shortcomings of microscopy, whether used as an auxiliary measure for microsurgery or alone with proficient endoscopic techniques, it will provide greater application in minimally invasive surgery for fourth ventricle and brainstem lesions. By utilizing the excellent degree of freedom of transcranial neuroendoscopy, there is no need to open the posterior arch of the atlas, making the surgery more minimally invasive. However, the sample size of this study is small, and it was completed under the very mature neuroendoscopic technology of our team. Its general safety and practicality still require extensive clinical research validation.

Introduction

Lesions located deep in the fourth ventricle and/or pontine tegmentum are challenges for neurosurgeons due to the limited working space and the complexity of the surrounding structures which include the medulla, lower cranial nerve nuclei, cerebellar peduncles, and posterior inferior cerebellar artery (PICA). Traditional surgery approach for this area involves splitting the inferior vermis to gain better direct access, which is also known as the transvermian approach. However, this approach can lead to postoperative disturbances of equilibrium and cerebellar mutism [ 1 ]. Operating through cerebellomedullary fissure by telovelar approach has also been used to reach this area without damaging inferior vermis which has been confirmed to be a reliable approach [ 2 , 3 , 4 ]. During the last 3 decades, all telovelar approaches have been performed under the microscope, and gradually found that it was difficult to access the upper fourth ventricle and pontine tegmentum [ 1 , 5 ]. To date, no case of upper fourth ventricle and pontine tegmentum lesions resected by full neuroendoscopic telovelar approach was reported. Our early clinical research shows that transcranial neuroendoscopic has obvious advantages in retrosigmoid approach, which can make up for the deficiency of microscopic surgery [ 6 ]. In this clinical study, we tentatively used the full transcranial neuroendoscopic telovelar approach to remove the lesions of brainstem and fourth ventricle, and most patients recovered excellent.

Data and method

General data.

In this study, the data of neurosurgery patients in our hospital from March 2020 to March 2022 were collected. The inclusion criteria were patients with treatment of brainstem and fourth ventricle lesions by the full neuroendoscopic in telovelar approach, including brainstem cavernous hemangioma, brainstem tumor, etc. The preoperative and postoperative imaging data, intraoperative neuroendoscopic video images and clinical manifestations were collected. A total of 5 patients with complete data were collected, including 1 male and 4 females, aged 3–67 years. The detailed clinical information is shown in Table 1 .

Surgical technique

Surgical procedures were performed with the patient in the lateral prone position, the head fixed in a head-holder with slight flexion. A midline suboccipital craniotomy was performed to expose the craniovertebral junction, and the posterior arch of the atlas was preserved. Under neuroendoscopic view, a Y-shaped dural opening was made, and the inferior edge of the tonsils, uvula, PICA and obex were exposed (Figs. 1 – 5 ). The tonsil and uvula were elevated and retracted by a thin transparent endoport, and then the tela choroidea, inferior medullary velum and floor of the fourth ventricle were visualized and protected. Looking forward to the upper ventricle, the hematoma was identified and removed under the neuroendoscopy. With further access to the cerebellar peduncle and pontine tegmentum area, the residual hematoma in the cerebellar peduncle was cleared away, and a small, cavernous malformation in the pontine tegmentum was identified and removed (Figs. 1 , 2 and 3 ).

Resection a middle cerebellar peduncle and pontine tegmentum cavernous malformation by full neuroendoscopic telovelar approach. A CT imaging shown a small hematoma in the right fourth ventricle, middle cerebellar peduncle and pontine tegmentum. B MRI SWAN sequence suggested a cavernous malformation; C patient in the lateral prone position with the head fixed in a head-holder and slightly flexed; D a midline suboccipital craniotomy was performed, and the posterior arch of C1 was preserved; E under neuroendoscopic view, the inferior edge of the tonsils, uvula, PICA and obex was exposed. F The tela choroidea, inferior medullary velum and floor of the fourth ventricle were visualized and protected. G The hematoma in the fourth ventricle was exposed and removed. H The hematoma in the middle cerebellar peduncle was exposed and removed. I A small, cavernous malformation in the pontine tegmentum was identified and removed. J The cavity was checked after the operation. K Postoperative CT scan shown the hematoma was removed completely. L Histopathological examination revealed a cavernous malformation

Resection a pontine tegmentum cavernous malformation by full neuroendoscopic telovelar approach. A CT imaging shown a small hematoma in the left pontine tegmentum. B T2-MRI shown irregular and heterogeneous signals in pontine tegmentum; C MRI SWAN sequence suggested a cavernous malformation; D head fixed in a head-holder and slightly flexed; E the median aperture of the fourth ventricle was exposed. F The tela choroidea, inferior medullary velum was visualized and protected. G The floor of the fourth ventricle was exposed. H The upper fourth ventricle floor was exposed, and the median eminence yellowing. I A small, cavernous malformation in the pontine tegmentum was identified and removed. J The patient recovered well after the operation. K Postoperative CT scan shown the hematoma was removed completely. L Histopathological examination revealed a cavernous malformation

Resection a right pontine arm cavernous malformation by full neuroendoscopic telovelar approach. A CT imaging shown a small hematoma in the right pontine arm. B – D MRI showed a cavernous hemangioma with hemorrhage in the right pontine arm. E Head fixed in a head-holder and slightly flexed. F The lesion was seen under neuroendoscopy. G Resection of lesion under neuroendoscopy. H , I After lesion resection, neuroendoscopic exploration of the fourth ventricle showed the outlet of aqueduct of midbrain. J The complete lesion. K Postoperative CT scan shown the lesion was removed completely. L Histopathological examination revealed a cavernous malformation

The posterior median foramen of the fourth ventricle was explored under neuroendoscopy. It was found that the tumor had protruded from the posterior median foramen. The important structures of the latch of medulla oblongata and brain stem were protected, and the tumor was removed in blocks (Fig.  4 ). Explored the floor of the fourth ventricle under neuroendoscopy. It was found that the lesion was located at the back of the pons. Blocked the outlet of the midbrain aqueduct, separated and removed the lesion. The exploration showed that the cerebrospinal fluid circulation of the fourth ventricle and midbrain aqueduct was unobstructed (Fig.  5 ).

Total neuroendoscopic resection of brainstem medulloblastoma. A – C MRI showed space occupying lesions of brainstem. D Head fixed in a head-holder and slightly flexed. E The lesion was seen under neuroendoscopy. F Resection of lesion under neuroendoscopy. G The lesion was completely resected under neuroendoscopy. H Postoperative CT scan shown the lesion was removed completely. I Histopathological examination revealed a medulloblastoma

Total neuroendoscopic resection of astrocytoma of brainstem and fourth ventricle. A – C MRI showed space occupying lesions in the brain stem and the top of the fourth ventricle. D MRI showed supratentorial hydrocephalus. E Head fixed in a head-holder and slightly flexed. F The median aperture of the fourth ventricle was exposed. G The lesion was seen under neuroendoscopy. H Resection of lesion under neuroendoscopy. I Enlarged midbrain aqueduct outlet after hydrocephalus. J Exploration of the third ventricle via midbrain aqueduct using neuroendoscopy. K: Postoperative CT scan shown the lesion was removed completely. L Histopathological examination revealed a pilocytic astrocytoma. SM: stria medullaris; VII: facial colliculus; XII: hypoglossal triangle; X: vagal triangle; GT: gracile tubercle; MA: median aperture; MS: median sulcus; ME: median eminence; AM: aqueduct of midbrain; PICA: posterior inferior cerebellar artery

Clinical presentation

A 62-year-old woman presented with dizziness and vomiting for 12 days and was transferred to our hospital. Neurological examination showed difficulty walking, and other exams were normal for cranial nerves, motor/sensory function, coordination, and reflexes. CT imaging showed small hematoma in the right fourth ventricle, middle cerebellar peduncle, and pontine tegmentum. Magnetic resonance imaging (MRI) performed with T1-weighted, T2-weighted and SWAN sequences suggested a cavernous malformation. The preoperative diagnosis was a cavernous malformation in the right middle cerebellar peduncle and pontine tegmentum. After discussion with her family, suboccipital craniotomy was planned.

A 54-year-old woman presented with numbness in right limb for one year. One year ago, she felt numbness in the right limb and CT scan shown small hemorrhage in the brainstem. After conservative treatment, she was improved and discharged from local hospital. But two weeks ago, she felt numbness in right side and came to hospital again, and CT scan shown rebleeding in the brainstem. Then she was admission to our department and MRI suggested a cavernous malformation in the left pontine tegmentum. After consent of the patient and her family, suboccipital craniotomy with telovelar approach by the neuroendoscopy was prepared.

A 67-year-old woman with dizziness for 2 months. Brain MRI examination in the local hospital found a lesion on the right side of brainstem, so she was referred to our hospital. Brain MRI + SWI in our hospital showed cavernous hemangioma of the right pontine arm. Communicate the condition and operation plan with the patient and her family, and operate with the consent. The patient recovered well after operation, the symptoms of dizziness disappeared, and there were no obvious complications after discharge.

A 3-year-old boy with unstable walking for 3 days. Before admission, the child was unstable in walking, fell for many times, and his body tilted to the right. Brain CT of the local hospital showed a lesion on the brainstem. After being transferred to our hospital, the child had rapid heart rate, disturbance of consciousness, vomiting and aspiration, airway obstruction, endotracheal intubation, and other rescue treatment. The emergency brain MRI + enhanced examination showed that the medulloblastoma on the brainstem were possible. The patient was in critical condition. After communicating with the family members and obtaining the consent of the family members, the emergency craniotomy was performed. After the operation, the child was complicated with severe pneumonia and transferred to pediatric ICU for further treatment. Pathological examination showed medulloblastoma. The child’s vital signs were unstable, and the prognosis was very poor. So the family chose to give up treatment and leave the hospital automatically.

A 32-year-old woman with dizziness and nausea for 3 months. The patient went to the local hospital and underwent brain MRI, which showed that the obstructive hydrocephalus. After referral to our hospital, brain MRI + enhanced examination showed that the lower end of midbrain aqueduct occupied space and the supratentorial hydrocephalus was formed. Communicate the condition and operation plan with patient and her families and operate with the consent. Pathological examination showed pilocytic astrocytoma. The patient recovered well after operation, the symptoms of dizziness and nausea disappeared, there were no obvious complications after discharge, and the hydrocephalus improved after follow-up.

Among the 5 patients, there were 3 cases of brainstem cavernous hemangioma and 2 cases of brainstem tumor (1 case of medulloblastoma, 1 case of pilocytic astrocytoma). All patients underwent neuroendoscopic telovelar approach to remove the lesions, and the operation was successfully completed. The 4 patients recovered well; the other 1 patient had serious complications of other systems after the operation and discharged automatically.

The symptoms of dizziness and vomiting disappeared immediately after the surgery in the Case 1, 3 and 5, and the numbness was decreased in the Case 2. These four patients’ hospital course and recovery were uneventful. They recovered well and showed no new signs of brainstem or cerebellar dysfunction and were discharged after 2 weeks postoperatively. The Case 4 was discharged automatically with serious complications of other systems after the operation.

Surgical access to lesions in the fourth ventricle may be achieved by utilizing transvermian or telovelar approach. Traditional transvermian approaches require splitting of the inferior vermis to gain better direct access from the posterior direction to the fourth ventricle. This approach inflicts the midline cerebellar structures and has been implicated in postoperative “cerebellar mutism syndromes” [ 3 ].

Matsushima et al. [ 7 ] firstly described the microsurgical anatomy of the cerebellomedullary fissure and found that it was actually a virtual space existing between the cerebellum and the medulla oblongata that was a natural corridor to the fourth ventricle. By elevating the cerebellar tonsils and opening the cerebellomedullary fissure, the tela choroidea and the inferior medullary velum could offer wide access to the fourth ventricle cavity without the need for splitting the vermis [ 8 ]. This technique was applied by other surgeons and has been developed and described as the “telovelar approach” [ 1 , 8 , 9 ]. The telovelar approach can widely expose the fourth ventricle from bottom to top by opening the choroid and inferior medullary sail, without cutting the cerebellar vermis, to reduce the occurrence of postoperative silent syndrome. When using the transparent sheath, we should pay attention to protect the structures of the latch of medulla oblongata and the floor of the fourth ventricle.

Since then, numerous reports of resection of various fourth ventricle tumors, arteriovenous malformations and aneurysms via this approach have been extensively described [ 1 , 3 , 9 , 10 ], and it appears that this approach has the potential to become the standard treatment for most lesions of the fourth ventricle with satisfactory results [ 1 ]. However, in terms of the vertical working angle of the microscope, it was easy to look from the roof to the floor of the fourth ventricle but was very difficult to assess from caudal to rostral via the telovelar approach. Therefore, this approach was limited in achieving access to the rostral third of the fourth ventricle and middle cerebellar peduncle. Under microscope, a possible maneuver to achieve more favorable working angle to the upper ventricle is to cut the posterior arch of the atlas. Deshmukh et al. found that additional removal of the C1 arch offered a larger working area that contributed to reaching the rostral half of the fourth ventricle [ 11 ]. However, it is difficult to access upper fourth ventricle from caudal to rostral without removal posterior arch of the atlas due to the vertical working angle of microscope. Neuroendoscopy has a good degree of freedom in surgery and can reach this area easily. Neuroendoscopy can overcome the limitations of microscope by taking advantage of the freedom achieved during surgery, as the neuroendoscopy can be operated easily from caudal to rostral in the fourth ventricle. Several authors have reported using angled endoscopy assistance for facilitating additional inspection around the anatomic corners and of tumor resection in the fourth ventricle [ 12 ], but none full neuroendoscopic telovelar approach was attempted. Our successful cases demonstrate that access to the upper fourth ventricle via the full neuroendoscopic telovelar approach without removing the posterior arch of the atlas is feasibility.

Conclusions

The telovelar approach has gained popularity as a safe and effective strategy for lesions in fourth ventricular and pons. However, without removing the posterior arch of the atlas, it is difficult to enter the upper part of the fourth ventricle under a microscope. Transcranial neuroendoscopy can effectively compensate for the shortcomings of microscopy, whether used as an auxiliary measure for microsurgery or alone with proficient endoscopic techniques, it will provide greater application in minimally invasive surgery for fourth ventricle and brainstem lesions. By utilizing the excellent degree of freedom of transcranial neuroendoscopy, there is no need to open the posterior arch of the atlas, making the surgery more minimally invasive. However, the sample size of this study is small, and it was completed under the very mature neuroendoscopic technology of our team. Its general safety and practicality still require extensive clinical research validation.

Availability of data and materials

All data meet the requirements of the journal and can be used by the magazine at will.

Tomasello F, Conti A, Cardali S, La Torre D, Angileri FF. Telovelar approach to fourth ventricle tumors: highlights and limitations. World Neurosurg. 2015;83:1141–7.

Article   PubMed   Google Scholar  

Tanriover N, Ulm AJ, Rhoton AL Jr, Yasuda A. Comparison of the transvermian and telovelar approaches to the fourth ventricle. J Neurosurg. 2004;101:484–98.

Jittapiromsak P, Sabuncuoglu H, Deshmukh P, Spetzler RF, Preul MC. Accessing the recesses of the fourth ventricle: comparison of tonsillar retraction and resection in the telovelar approach. Neurosurgery. 2010;66:30–9.

PubMed   Google Scholar  

Liu R, Kasper EM. Bilateral telovelar approach: a safe route revisited for resections of various large fourth ventricle tumors. Surg Neurol Int. 2014;5:16.

Article   PubMed   PubMed Central   Google Scholar  

Han S, Wang Z, Wang Y, Wu A. Transcerebellomedullary fissure approach to lesions of the fourth ventricle: less is more? Acta Neurochir (Wien). 2013;155:1011–6.

Cai Q, Li Z, Guo Q, Wenju W, Baowei J, Zhibiao C, Hongjuan D, Shanping M. Microvascular decompression using a fully transcranial neuroendoscopic approach. Br J Neurosurg. 2021. https://doi.org/10.1080/02688697.2020.1820943 .

Matsushima T, Fukui M, Inoue T, Natori Y, Baba T, Fujii K. Microsurgical and magnetic resonance imaging anatomy of the cerebello-medullary fissure and its application during fourth ventricle surgery. Neurosurgery. 1992;30:325–30.

Matsushima T, Abe H, Kawashima M, Inoue T. Exposure of the wide interior of the fourth ventricle without splitting the vermis: importance of cutting procedures for the tela choroidea. Neurosurg Rev. 2012;35:563–71 ( discussion 571–562 ).

Ziyal IM, Sekhar LN, Salas E. Subtonsillar-transcerebellomedullary approach to lesions involving the fourth ventricle, the cerebellomedullary fissure and the lateral brainstem. Br J Neurosurg. 1999;13:276–84.

Shigeno T, Kumai J, Endo M, Hotta S. Surgery of AVM of the inferior medullary velum by the uvulotonsillar approach–advantage of moving of the cerebellar tonsil: technical case report. No Shinkei Geka. 2002;30:87–92.

Deshmukh VR, Figueiredo EG, Deshmukh P, Crawford NR, Preul MC, Spetzler RF. Quantification and comparison of telovelar and transvermian approaches to the fourth ventricle. Neurosurgery. 2006;58:ONS 202-206 ( discussion ONS-206-207 ).

Article   Google Scholar  

Ghali MGZ. Telovelar surgical approach. Neurosurg Rev. 2021;44(1):61–76.

Download references

This work was supported by National Natural Science Foundation of China (81671306; 81971158), Wuhan Science and Technology project (2019020701011470).

Author information

Long Zhou, Hangyu Wei and Zhiyang Li are co-first authors.

Authors and Affiliations

Department of Neurosurgery, Renmin Hospital of Wuhan University, No. 238, Jiefang Road, Wuchang District, Wuhan City, 430060, Hubei Province, China

Long Zhou, Hangyu Wei, Zhiyang Li, Huikai Zhang, Ping Song, Wenju Wang, Pan Lei, Qianxue Chen, Zaiming Liu, Hui Ye & Qiang Cai

Department of Critical Care Medicine, Eastern Campus, Renmin Hospital of Wuhan University, Wuhan, China

Department of Neurosurgery, Xiantao First People’s Hospital of Yangtze University, No. 29, Middle Part of Mianzhou Avenue, Xiantao City, 433000, Hubei Province, China

You can also search for this author in PubMed   Google Scholar

Contributions

QC and DS study concept and design, critical revision of manuscript for intellectual content, acquisition of data. LZ, HW and ZL collect and analysis data, and write manuscripts. HZ, PS, LC, WW, PL, QC, ZL, HY collection and interpretation of data.

Corresponding authors

Correspondence to Daofa Sun or Qiang Cai .

Ethics declarations

Ethics approval and consent to participate.

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. And it has been approved by the Ethics Committee of Clinical Research, Renmin Hospital of Wuhan University (WDRY2022-K098).

Consent for publication

Informed consent was obtained from all individual participants included in the study.

Competing interests

The authors claim that none of the material in the paper has been published or is under consideration for publication elsewhere, and no competing interests were disclosed.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Cite this article.

Zhou, L., Wei, H., Li, Z. et al. Treatment of brainstem and fourth ventricle lesions by the full neuroendoscopic telovelar approach. Eur J Med Res 28 , 564 (2023). https://doi.org/10.1186/s40001-023-01460-5

Download citation

Received : 25 September 2022

Accepted : 19 October 2023

Published : 05 December 2023

DOI : https://doi.org/10.1186/s40001-023-01460-5

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Transcranial neuroendoscope
• Telovelar approach
• Brainstem lesion
• Fourth ventricle
• Cavernous hemangioma

European Journal of Medical Research

ISSN: 2047-783X

• Submission enquiries: [email protected]
• General enquiries: [email protected]

This paper is in the following e-collection/theme issue:

Published on 4.12.2023 in Vol 25 (2023)

Effectiveness of Different Telerehabilitation Strategies on Pain and Physical Function in Patients With Knee Osteoarthritis: Systematic Review and Meta-Analysis

Authors of this article:

Wu Xiang   1, 2 * , MM ;   Jun-Yu Wang   2 * , PhD ;   Bing-jin Ji   1 , BM ;   Li-Jun Li   1 , MM ;   Han Xiang   3 , MM

1 Department of Rehabilitation, Beibei Traditional Chinese Medical Hospital, Chongqing, China

2 Department of Rehabilitation Medicine, Shanghai Fourth People’s Hospital Affiliated to Tongji University School of Medicine, Shanghai, China

3 Department of Radiology, Daping Hospital, Army Medical University, Chongqing, China

*these authors contributed equally

Corresponding Author:

• Han Xiang , MM
• Department of Radiology
• Daping Hospital
• Army Medical University
• Yuzhong District
• No. 10 Changjiang Branch Road
• Chongqing , 400042
• Phone: 86 23 6874 6920
• Email: [email protected]

• IDX Channel
• Motion Insure
• Motion Trade
• Motion Banking
• Mister Aladin
• Asahan Raya
• Bengkulu Utara
• Lintasbabel
• Lhokseumawe
• Portal Aceh
• Bandung Raya
• Banjarnegara
• Ciamis Raya
• Jateng Info
• Joglo Semar
• Karanganyar
• Pangandaran
• Probolinggo
• Tasikmalaya
• Tulungagung
• Barito Info
• Sorong Raya
• Detail Berita

Contoh Jurnal Ilmiah Beserta Cara Membuatnya yang Baik dan Benar

JAKARTA, iNews.id - Contoh jurnal ilmiah berikut ini bisa menjadi referensi bagi kamu yang hendak membuat jurnal. Jurnal ilmiah atau jurnal akademik merupakan publikasi ilmiah yang memuat kumpulan artikel ilmiah yang kemudian dipublikasikan secara reguler.

Tujuan penulisan jurnal ilmiah adalah untuk mengembangkan sebuah penelitian yang sudah ditulis, serta menjadi acuan bagi peneliti lain yang melakukan penelitian sejenis.

Tak hanya itu, jurnal ilmiah juga digunakan sebagai media untuk mengembangkan suatu bidang keilmuan dari sejumlah penelitian yang telah dilakukan.

Jurnal ilmiah memuat judul, abstrak, pendahuluan, pembahasan atau analisis, kesimpulan, dan daftar pustaka.   Agar dapat lebih memahaminya, kamu bisa menyimak contoh jurnal ilmiah beserta cara membuatnya berikut ini.

Contoh Jurnal Ilmiah

Sebelum membahas tentang contoh jurnal, ada baiknya kamu mengetahui cara membuat jurnal ilmiah terlebih dahulu. Cara membuat jurnal ilmiah sendiri mirip dengan proses pembuatan skripsi, namun lebih sederhana. Berikut ini cara membuat jurnal ilmiah yang perlu kamu ketahui.

1. Judul Jurnal

Dikutip dari pskp.kemdikbud.go.id, judul ini disusun dengan menggambarkan isi tulisan secara ringkas namun jelas, serta mampu menarik minat orang lain untuk membaca. Dalam menulis judul, penulis juga tidak boleh memberikan makna ganda (ambigu). 

Penulisan judul jurnal ini berada di tengah atas halaman, menggunakan huruf kapital, dan dicetak tebal. Kemudian disarankan membuat judul tidak melebihi 12 kata jika jurnal ini berbahasa Indonesia, atau 10 kata jika ditulis dalam bahasa Inggris.

Cara membuat jurnal selanjutnya adalah dengan mencantumkan nama penulis, nama pembimbing, dan juga nama lembaga jika tanpa gelar akademik.

Abstrak merupakan bagian dalam jurnal ilmiah yang berfungsi untuk menggambarkan detail jurnal yang disusun secara padat. Abstrak biasanya berisi latar belakang penelitian, tujuan penelitian, metodologi dan hasil penelitian. Abstrak ditulis dengan panjang antara 150-200 kata.

4. Pendahuluan

Pendahuluan merupakan bagian yang menguraikan latar belakang permasalahan yang ingin dikaji atau ingin dibahas. Selain itu, sertakan juga manfaat dan ulasan dari jurnal agar pembaca mengetahui tujuan jurnal tersebut.

5. Metodologi

Bagian metodologi ini menjelaskan objek atau subjek kajian, variabel penelitian serta lokasi pengambilan data. Selain itu, dijelaskan pula metode analisis dan pengolahan data.

7. Hasil dan Pembahasan

Dalam penyajian hasil penelitian, penulis perlu menyajikan data tersebut apa adanya, tanpa interpretasi atau pendapat subjektif. Data disajikan dengan ringkas dalam bentuk teks, tabel, maupun gambar.

8. Kesimpulan

Kunci dalam membuat kesimpulan adalah peneliti mengacu pada pernyataan ketika menyajikan informasi apapun yang dipelajari. Dalam bagian kesimpulan, penulis harus menjelaskan pembuktian hipotesis yang dibahas dalam pendahuluan dengan temuan dalam hasil pembahasan.

Kesimpulan ini harus mampu menerangkan keterkaitan kedua hal tersebut yang menjadi tujuan utama dalam pembuatan jurnal ilmiah.

9. Daftar Pustaka

Daftar pustaka merupakan bagian akhir dalam jurnal ilmiah. Bagian ini berisi buku atau jurnal yang menjadi bahan rujukan dalam proses pembuatan jurnal.

Editor : Komaruddin Bagja

Follow Berita iNews di Google News

BERITA TERKAIT

8 Cara Membuat Jurnal yang baik dan Benar sesuai Kaidah untuk Referensi

Target Parpol Pendukung Ganjar Pranowo 7 Bulan Sebelum Pilpres 2024

Berita di Sekitarmu

Lokasi Tidak Terdeteksi

Aktifkan fitur berita di sekitar Anda?

Pastikan pengaturan lokasi browser Anda aktif

Non-aktifkan fitur berita di sekitar Anda?

Ketahui lebih lanjut