A significant challenge for psychiatry is to explain precisely how the brain generates psychopathology, as its translation is presumed to advance effective mechanism-based treatments. Computational psychiatry – a mathematical understanding of mental illness – has emerged to bridge this explanatory gap [1]. Broadly, computational psychiatry uses mathematical models to study psychiatric disorders, typically done via 1) an explanatory quantitative modelling approach to explain how aberrant computations of the mind produce psychiatric symptoms, and 2) data-driven modelling, commonly used to predict and track symptom progression. These methods have been applied to identify clinically relevant markers in psychiatry [2,3,4]. Recently, start-ups have been applying these principles to clinical settings for aiding diagnosis (e.g., https://limbic.ai/) and delivering personalised psychotherapy (e.g., https://alena.com/).
Early career researchers (ECRs) are uniquely positioned to advance the translation of computational psychiatry. However, during our own academic training, we encountered barriers that may limit its uptake amongst ECRs. Here, we highlight these barriers and propose potential solutions.
Accessibility
Most mental health researchers are not mathematically trained. Computational psychiatry requires foundational knowledge in statistical modelling and programming for its application or understanding. However, such training is not routinely embedded into postgraduate programs. Most computational psychiatry studies are conducted by institutions with strong computational foundations and training capacity, often within Western, Educated, Industrial, Rich, Democratic (WEIRD) settings.
To address accessibility, we propose three potential solutions. One, mental health research labs should consider inter-disciplinary collaborations with researchers who have necessary quantitative skills (e.g., data scientists, computer scientists) to manage complex data. This would not only promote mutual learning and crosstalk between psychiatry, psychology, and computational neuroscience, but also enhance research efficiency (i.e., obviating the need to learn complex analyses from scratch). Two, upskilling ECRs in quantitative skills should be prioritised. Universities could embed programming and quantitative courses into psychology-based programs to cater for those interested in pursuing computational psychiatry (e.g., undergraduate subjects on computational modelling, MSc in computational cognitive science). Code-sharing alongside publications should be encouraged as they are invaluable learning resources for ECRs. Three, upskilling ECRs should involve researchers from the Global South, as the development of translatable computational models should not ignore cross-cultural validity. High-income countries could collaborate, or initiate specific schemes to encourage sharing of knowledge and resources with the Global South (e.g. the Oxford University Wellcome Centre Integrative Neuroimaging Global Scholars (WINGS) programme for researchers from countries with limited research investment).
Translatability
Although translational computational psychiatry seems promising, there are still significant ways to go. One notable limitation is the poor psychometric properties of computational parameters, such as limited construct validity (the extent to which a test measures the intended construct) and test-retest reliability (rate of agreement between measurements taken from the same individual at different timepoints) [5]. These problems significantly impede the reliability of computational parameters as markers for psychiatry, which in part explains why these parameters are not incorporated in routine clinical practice [6]. Furthermore, although psychiatry research recognizes sociocultural factors as determinants of mental ill health, present work in the field rarely incorporates these factors into computational models of mental health [7]. Indeed, accounting for sociocultural considerations has proven successful in adjacent fields like clinical psychology (e.g., incorporating ethnicity and neighbourhood disadvantage into models better explains psychological well-being [8] and psychiatric service usage [9]).
To enhance translatability, we must improve the psychometric properties of computational psychiatry models. Computational findings must always be validated with simulations and parameter recovery [10]. Next, clinically-trained ECRs could also administer behavioural paradigms alongside clinical care to verify the relevance of computational constructs to symptomatology and treatment responses, thereby accelerating their uptake in clinical settings. Similarly, ECRs should be encouraged to collaborate with healthcare sectors, start-ups, and think-tanks to enhance the policy relevance of their research. Academic-industry partnerships could be incorporated into PhD training programs, e.g., the Medical Research Council’s (MRC) industrial Collaborative Awards in Science and Engineering (iCASE) studentship that includes brief collaborations with non-academic partners.
Simultaneously, incorporating patient-public involvement (PPI) in computational psychiatry research can ensure that outcomes have clear clinical relevance from conception. PPI’s benefits are three-fold: 1) it can better inform models or tasks to enhance specificity; 2) it serves as a platform to check model assumptions in ethnically and socioeconomically diverse patient groups; 3) it allows for examination of whether tasks are meaningful and comprehensible to patients. As an example for point 3, the well-known two-step reinforcement learning task, conceived as computational proxies for goal-directed actions and habits in psychiatry [11], might be over-interpreted as it is highly instruction-dependent [12]. Indeed, almost all participants became more goal-directed/model-based when instructions were made more explicit [12] – something that may have been detected earlier with PPI. To facilitate PPI efforts, we advocate for training ECRs in science communication and outreach early on [13, 14], equipping them with skills necessary to communicate research effectively to non-specialist PPI panels.
Funding
Psychiatry is complex and dynamic. For computational psychiatry to be informative (e.g., identifying markers), there is a need for longitudinal studies with large samples and multi-disciplinary collaborations. Multi-site consortia-based research (e.g., IMAGEN, ENIGMA, and ABCD) have proven that large collaborations over longer timescales can be successful, albeit requires substantial funding. Existing major funders of psychiatry research often offer fellowships and grants for shorter timescales (i.e., 2–5 years), and are highly competitive, which may inadvertently promote “quantity over quality” practices to maintain competitiveness.
High quality and translatable computational psychiatry work requires slow rigorous research, as developing robust paradigms and meaningful models of mental health incurs significant time and resources. Funders should promote ‘slow science’, emphasising quality over quantity and allow for greater timescales for research that advances the field [15], e.g., more longitudinal research. To support ECRs, funding bodies could consider 1) moving away from publication-based milestones and instead emphasise clinical and societal impact, and 2) providing structured mentorship opportunities with established computational psychiatrists.
In conclusion, ECRs form the future for computational psychiatry, but ensuring high-quality work requires support from labs, research institutions, and funders. To play our part, we have curated a list of resources for ECRs interested in delving into computational psychiatry.
References
Huys QJM, Maia TV, Frank MJ. Computational psychiatry as a bridge from neuroscience to clinical applications. Nat Neurosci. 2016;19:404–13.
Webb CA, Trivedi MH, Cohen ZD, Dillon DG, Fournier JC, Goer F, et al. Personalized prediction of antidepressant v. placebo response: evidence from the EMBARC study. Psychological Med. 2019;49:1118–27.
Berwian IM, Wenzel JG, Collins AGE, Seifritz E, Stephan KE, Walter H, et al. Computational Mechanisms of Effort and Reward Decisions in Patients With Depression and Their Association With Relapse After Antidepressant Discontinuation. JAMA Psychiatry. 2020;77:513.
Konova AB, Lopez-Guzman S, Urmanche A, Ross S, Louie K, Rotrosen J, et al. Computational Markers of Risky Decision-making for Identification of Temporal Windows of Vulnerability to Opioid Use in a Real-world Clinical Setting. JAMA Psychiatry. 2020;77:368–77.
Karvelis P, Paulus MP, Diaconescu AO. Individual differences in computational psychiatry: A review of current challenges. Neurosci Biobehav Rev. 2023;148:105137.
Yip SW, Barch DM, Chase HW, Flagel S, Huys QJM, Konova AB, et al. From Computation to Clinic. Biol Psychiatry Glob Open Sci. 2023;3:319–28.
Hitchcock PF, Fried EI, Frank MJ. Computational Psychiatry Needs Time and Context. Annu Rev Psychol. 2022;73:243–70.
Alamilla SG, Scott MA, Hughes DL. The Relationship of Individual- and Community-Level Sociocultural and Neighborhood Factors to the Mental Health of Ethnic Groups in Two Large U.s. Cities. J Community Psychol. 2016;44:51–77.
Hansen MC, Aranda MP. Sociocultural influences on mental health service use by Latino older adults for emotional distress: Exploring the mediating and moderating role of informal social support. Soc Sci Med. 2012;75:2134–42.
Wilson RC, Collins AG. Ten simple rules for the computational modeling of behavioral data. eLife. 2019;8:e49547.
Voon V, Reiter A, Sebold M, Groman S. Model-Based Control in Dimensional Psychiatry. Biol Psychiatry. 2017;82:391–400.
Feher da Silva C, Hare TA. Humans primarily use model-based inference in the two-stage task. Nat Hum Behav. 2020;4:1053–66.
Arumugam A, Phillips LR, Moore A, Kumaran SD, Sampath KK, Migliorini F, et al. Patient and public involvement in research: a review of practical resources for young investigators. BMC Rheumatol. 2023;7:2.
Rouzer SK, Kalinowski LM, Kaseda ET. The importance of promoting scientific advocacy & outreach for trainees. Neuropsychopharmacol. 2023;48:713–5.
Frith U. Fast Lane to Slow Science. Trends Cogn Sci. 2020;24:1–2.
Acknowledgements
The authors would like to thank the NPP Special Projects Team for providing the opportunity and the platform to voice their ideas. The authors are grateful to Dorothy Langton and her family for their support.
Funding
All research at the Department of Psychiatry in the University of Cambridge was supported by the NIHR Cambridge Biomedical Research Centre (NIHR203312) and the NIHR Applied Research Collaboration East of England. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.
Author information
Authors and Affiliations
Contributions
AAM and TVL conceptualised, gathered the resources for, wrote, and revised the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
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/.
About this article
Cite this article
Marzuki, A.A., Lim, T.V. Bridging minds and policies: supporting early career researchers in translating computational psychiatry research. Neuropsychopharmacol. 49, 903–904 (2024). https://doi.org/10.1038/s41386-024-01834-1
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41386-024-01834-1