1/7-PsycheMERGE: Advancing Precision Psychiatry

PROJECT SUMMARY Psychiatric disorders remain a leading cause of disability in the US and are associated with increased morbidity and mortality. Early detection and treatment is essential to improving long-term outcomes, yet a substantial proportion of patients with psychiatric complaints experience long diagnostic odysseys before receiving an

appropriate diagnosis and initiating effective treatment. “Learning health care systems” aim to short-circuit this slow process by leveraging the diagnostic, treatment, and utilization patterns left behind in “big data” (e.g., clinical, genomic, and social determinants of health) to more efficiently and accurately match the right patient

with the right diagnosis/treatment, at the right time. Furthermore, over the past several years, a new paradigm– precision medicine–has moved to the forefront of biomedical research and clinical practice. Precision medicine has been defined as “an approach to disease treatment and prevention that seeks to maximize effectiveness by

taking into account individual variability in genes, environment, and lifestyle.” Since its inception in 2018, the mission of the PsycheMERGE network has been to advance precision psychiatry in a learning health care system framework. This application, which was developed collaboratively by PsycheMERGE Network members,

represents an opportunity for profound advancement of both basic and translational research in precision psychiatry. We propose extending our foundational efforts to now address barriers to scalability, utility of genomic data, clinical application, and translation to clinical practice in a precision psychiatry paradigm. Specifically, Aim

1 creates a nation-wide federated transfer-learning platform for the development of generalizable and bias-aware algorithms. Aim 2 integrates state-of-the-art methods to perform inclusive trans-ancestry genomic analysis of biobank samples and further innovates by leveraging the breadth and depth of medical record data to discover

novel biology that can further inform precision psychiatry paradigms. Aim 3 addresses the application of algorithms by focusing on two use cases including (a) differential diagnosis between bipolar disorder 1 and other mood disorders, as well as (b) probabilistic treatment response to antidepressants for acute depressive

episodes. Lastly, Aim 4 uses mixed methods to assess the feasibility, utility, and attitudes towards precision psychiatry tools. Our combined sample of clinical EHR data exceeds 29 million individuals and of those, nearly 2 million also have genetic data already available for analysis across the twelve sites included in this application.

A cross-cutting theme throughout the application is the intentional focus on equitable performance of algorithms, innovative integration of social determinants of health, and inclusive methods for genomic analyses. The sites included are also representative of many diverse communities across the United States including the East and

West Coasts, the South, and the Midwest. This application represents a major step towards equitable precision psychiatry and brings the field closer to the goals outlined in the updated NIMH Strategic plan.