Computational Models of Mental Health Disorders: A Review

Abstract

Background 

Computational models are becoming an indispensable tool for studying and understanding the underlying mechanisms of various scientific phenomena. Thus, in recent years, the complexity of diagnosing and treating mental health disorders has proven to be an exciting challenge for the interdisciplinary field of computational psychiatry. Simply defined, this field represents the application of computational modeling and theoretical approaches to psychiatric questions (e.g., to explain the underlying mechanisms of psychopathologies) [1]. This review attempts to present, compare, and evaluate different approaches proposed so far in this field, both in general and in the case of one specific disorder (depression). 

Aims

This thesis has three main aims: (1) to provide a detailed and up-to-date review of computational models used in computational psychiatry (data-driven, theory-driven and combined approaches) [2], as well as other alternative models; (2) to synthesize findings from the application of these approaches to depression and to identify possible research gaps and under-studied phenomena. This particularly applies to the modeling approach to brain stimulation techniques, which are less commonly used in psychiatry than psychotherapy or pharmacology. For example, TMS (transcranial magnetic stimulation) is often used in cases of treatment-resistant depression.; (3) to offer a systematic overview of all the types of data used in computational psychiatry (neuroimaging, genetic, mobile (devices), behavioral data) and highlight possible problems that may be encountered when attempting to include them in models and operationalize them as meaningful constructs, often at different levels of analysis (e.g., relating neuroimaging and behavioral data). 

Methods and Results

Since this is a fairly new field, the research will be focused on the articles and books published from 2007 onwards (PubMed, ResearchGate, Google Scholar), as well as on material presented at the Computational Psychiatry Course Zurich (which is keeping up-to-date with the latest research in the field). An extensive literature search and review of articles, books and lectures from the field of computational psychiatry should provide a systematic review of the approaches applied to mental health disorders, identify possible research gaps, and highlight possible novel areas of application.

Conclusion

The goal of this review is to evaluate and compare the most suitable computational approaches to the study of mental health disorders and different dimensions of depression. The research also aims to contribute to a more systematic overview of data types used in this field, to facilitate their choice and implementation in computational models.

The hope is that novel insights gained from various computational approaches will contribute to the future development of translational and precision psychiatry, whose ultimate aim is to optimize treatments in clinical practice.

References

[1] P. Seriès, Computational Psychiatry: A Primer. Cambridge (Mass.): The MIT Press, 2020.

[2] Q. J. Huys, T. V. Maia, and M. J. Frank, “Computational psychiatry as a bridge from neuroscience to clinical applications,” Nature Neuroscience, vol. 19, no. 3, pp. 404–413, Feb. 2016. doi:10.1038/nn.4238