Health Informatics Seminar Series: Examining the Landscape of Human-Centered AI for Health Systems and Implications for Risk Prediction

Series Overview

The joint Health Informatics seminar series is a collaboration between the Faculty of Information (iSchool) and the Institute of Health Policy, Management and Evaluation. It aims to build bridges between people from these units, who share an interest in and passion for information and communication technology (ICT) for health—reflecting the interdisciplinary nature of Health Informatics, which is at the intersection of information, technology, social sciences, and health. Speakers will discuss their cutting-edge research on a variety of topics.

Session Description

As computational power and access to data-driven modeling becomes increasingly prevalent in healthcare settings, artificial intelligence (AI) and predictive models are being more frequently integrated into decision-making, risk prediction, and diagnosis tasks. These models can produce biased, unjust results, leading to unintended consequences and placing individuals at a disadvantage when seeking care. As well, while AI models are often promoted as “cost-effective alternatives” aiming to reduce health system spending, the computational, societal, and environmental costs of developing and deploying these models are often not considered. In this seminar we will explore the landscape of medical and clinical AI models as well as the granularity with which computational costs are acknowledged. We will also discuss current work examining the development of predictive models for diabetes care in Ontario.

Speaker Bios

Victoria Chui

Victoria is a PhD student at the Faculty of Information, University of Toronto. She investigates the development and usage of algorithms in medical, clinical, and administrative health system settings. In collaboration with the Dalla Lana School for Public Health, Victoria is currently examining how predictive risk models can be used in Ontario to predict type 2 diabetes onset and complications. She holds a Master of Information from the University of Toronto, where she previously researched the usage of time series models to predict patient volumes in surgical wards.

Shion Guha

Dr. Shion Guha is an Assistant Professor with the Faculty of Information with a cross-appointment to the Department of Computer Science, University of Toronto. He directs the Human-Centered Data Science Lab and is part of the broader Critical Computing research community, while also being affiliated with the Data Sciences Institute and Schwartz Reisman Institute. Dr. Guha’s work aims to integrate technical, computational approaches with critical, interpretive inquiry to holistically address pressing issues in public interest technology and policy.