Development of a deep learning framework to predict suicidal behavior, repeat hospitalization, and emergency department visits in mental health patients

The Canadian Institute for Health Information (CIHI) is a not-for-profit organization that provides data standards and collection, analytical reports, and analytical tools to the Canadian health sector. CIHI has a five to ten year mandate to prioritize analytics for vulnerable populations, including individuals suffering from mental illness. Advances in deep learning and the availability of large health datasets, broadly and in some cases exclusively available to CIHI, has paved the way for new approaches to health-related predictive analytics. As part of a postdoctoral fellowship at CIHI we propose to develop a deep learning-based framework that provides likelihood predictions for three future events in individuals with mental illness: 1) Probability of re-hospitalization, 2) Probability of emergency department visit, and 3) Probability of suicidal behavior. These predictions can help identify at-risk populations in order to adjust health services, reduce exposure to modifiable risk factors, evaluate suitability for patient discharge, provide early intervention, inform policy, and provide guidance for funding decisions based on predicted demands on mental health services. In addition to my work at CIHI, I will be part of the Ottawa Hospital Research Institute (OHRI) and the Department of Medicine at the University of Ottawa under the supervision of Dr. Peter Tanuseputro. The OHRI is the research branch of the Ottawa Hospital, which is one of the largest learning and research hospitals in Canada. With Dr. Tanuseputro I will be using health data to study the effects of physician mental health on patient outcome, with a focus on machine learning. This is a first-of-its-kind opportunity where patient health data can be linked to the treating physician and the physician's mental health. This project will help identify areas requiring attention in the well-being of physicians, and translates well to workplace mental health in general and during times such as COVID-19.