Grant: CIHR Fellowship - PA: PPHE Research - General

Depression is a mental health condition that affects millions of people worldwide, impacting individuals and society. Identifying depression early is essential for effective treatment, but traditional methods, such as self-reported questionnaires or assessments by healthcare professionals, can have limitations. People may be unaware of their symptoms or unintentionally mask their behaviours when interacting with clinicians. This research project seeks to create a more objective way of detecting depression using artificial intelligence (AI) techniques. Specifically, we will develop a model that uses natural language processing (NLP) to analyze speech, along with audio and video recordings, to detect signs of depression. By analyzing impromptu speech samples (when people speak freely without preparation) and the tone of voice and facial expressions, the model can potentially identify patterns associated with depression. We will collect data from 1,000 participants using an online platform called Prolific, which allows us to gather speech, audio, and video samples remotely. Participants will also complete a questionnaire that measures symptoms of depression. We will then train our model to analyze language, sound, and visual cues and screen for depression. We will also explore how demographic factors like age and sex influence these patterns, ensuring that the model is sensitive to differences across diverse populations. By combining information from speech, audio, and video, this research aims to improve how we detect depression. The outcomes could provide a valuable tool for early screening, helping individuals get timely support and contributing to addressing mental health challenges in Canada and beyond. With guidance from experts in psychiatry and engineering, and in collaboration with mental health professionals, linguists, and data scientists, this research could lead to significant advancements in how depression is diagnosed and treated.