Messy measurements…and the opportunities they provide in mental health
In the field of mental health, the vast majority of measurements are intrinsically "messy". This messiness stems from the difficulty in measuring ground-truth phenomena which are inaccessible. This results in off-target variations due to human interpretations, difficult-to-characterize biases, and widespread infrastructure inefficiencies. However, rather than viewing these factors solely as a hindrance, this talk explores the distinct opportunities they present to improve research robustness in research and clinical care. Drawing on concrete examples across MRI, biometrics, and clinical assessments, this presentation highlights how robust computational infrastructure can turn analytical variation into a tool for improving the generalizability of brain-behavior models. It also demonstrates how engaging the broader data science community through open data initiatives and competitions can help solve complex challenges, such as accurately detecting sleep states from wrist-worn actigraphy data. Finally, the talk illustrates how leveraging open data, software engineering, and carefully constrained Artificial Intelligence tools can transform unstructured clinical assessments into scalable, accessible tools for public good. Ultimately, this presentation outlines a principled framework for using software engineering and AI in mental health research and care in a way that can accelerate discovery and increase access without increasing the risk of harm.
Zoom link
https://mcgill.zoom.us/j/4937377259
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
Loading virtual attendance info...
- 3775 Rue University
- Montreal, Quebec
- Canada
- Building: Duff Medical Building
- Room Number: 321
- Starts 30 May 2026 04:00 AM UTC
- Ends 08 July 2026 03:00 PM UTC
- No Admission Charge
Speakers
Messy measurements…and the opportunities they provide in mental health
Biography:
Gregory Kiar, PhD, is the director of the Center for Data Analytics, Innovation, and Rigor (DAIR) at the Child Mind Institute. With a background in biomedical engineering, Greg has developed techniques to study many different forms of biosignal and health data, ranging from building turnkey tools to estimate brain connectivity maps and assess the robustness of neuroscience findings, to developing algorithms that assess personal risk from text data in journaling or electronic health records. He is a long-time advocate of open science, enthusiastic mentor, and constant seeker of interdisciplinary collaboration.
