Grant: Health System Impact Fellowships - PostDoc Central Canada 2 year

The Major Depressive Disorder Care Pathway (MDD-ICP) is a standardized pathway of care for clients with depression at the Centre for Addiction and Mental Health (CAMH). In the MDD-ICP, psychiatrists write detailed notes about their meetings with clients. Analyzing these notes can help improve our understanding of clients with depression and inform decisions about their care. However, these notes are challenging to analyze because they are unstructured. Advancements in language modelling and machine learning offer solutions to this issue. In the first stage of the Health Systems Impact Fellowship (HSIF), we will use these methods to analyze the MDD-ICP clinical notes. Specifically, we will predict clients' responses to treatment using machine learning models. We will train our models based on clinical notes from intake assessments, as well as questionnaire data available in CAMH health records. According to previous research, the ability of our models to predict treatment responses may depend on client demographics (like sex or income), as well as the writing style of psychiatrists. We will also assess for these biases in our models, by stratifying our analyses according to these characteristics. In the second stage of the HSIF, we will use our findings to develop support tools and workshops that help psychiatrists make clinical decisions and improve the quality of their notes. We will implement these interventions in small groups of MDD-ICP psychiatrists. To evaluate whether the interventions are useful, we will compare outcomes from these psychiatrists to groups of control psychiatrists. This HSIF advances a goal of CAMH's Krembil Centre for Neuroinformatics to gain insights into mental health using innovative technologies. By using these insights to inform clinical decisions, it addresses the MDD-ICP's aim to optimize treatments for clients with depression.