Evaluation of the Use of Longitudinal Data for Depression Research and Antidepressant Drug Discovery

Evaluation of the Use of Longitudinal Data for Depression Research and Antidepressant Drug Discovery

Alvin Sherman Library

Depression is one of the most prevalent mental health disorders, affecting over 280 million people globally. Diagnosis of depression remains limited, and treatment is complicated due to the subjectivity of using psychometric scales for diagnosis. This has created a need for more objective methods for the diagnosis of depression and the assessment of the outcome of various treatment/management methods. These methods should consider a holistic approach encompassing biochemical, mental, and socioeconomic factors of patients. Longitudinal datasets, in the age of AI, offer an unprecedented opportunity for such an approach. Thus, it is hypothesized that the application of machine learning to longitudinal datasets pertaining to menthal health can be an effective method for identifying novel biomarkers and/or optimizing suitable treatment options for stratified patient groups and individuals. We identified and assessed 55 major longitudinal datasets from around the world pertaining to mental health using Landscaping International Longitudinal Databases, including UkBioBank, AllOfUs, Netherlands Study of Depression and Anxiety, and more, as well as 2 identified datasets, Finngen and Million Veterans Program. The assessment criteria for datasets were based on the types of biomarkers available, i.e., genetic data (genotype/whole genome sequencing), mental health questionnaires, lifestyle questionnaires, and imaging data, as well as their linkage to electronic health records, activity monitoring, and biochemical laboratory tests. We also performed a PubMed literature search to identify research publications that have used these databases as part of their research. Overall, UkBioBank was the database most frequently used in depression research as indicated by the number of PubMed citations. AllOfUs was among the datasets with the most availability for biomarkers, suggesting a high potential for its use in depression diagnosis and treatment. Future work includes using machine learning methods in order to perform data analysis using the AllOfUs database.