Artificial intelligence (AI) is transforming health care, offering powerful tools for biomedical data analysis and clinical decision-making. This session will provide an overview of recent advances in applying state-of-the-art AI methods, including transformer-based foundation models, to medical image analysis. While these models have shown promise, significant challenges remain such as bias, limited interpretability and difficulty adapting to real-world clinical contexts.
Dr. Xiaoxiao Li, an assistant professor in the Department of Electrical and Computer Engineering at the University of British Columbia, a faculty member at Vector Institute and a visiting research scholar at Google, will discuss strategies to address these challenges with a focus on improving fairness, robustness and clinical utility. The talk will also examine the ongoing debate over the use of general-purpose foundation models compared with specialized medical models, considering their cost, scalability and impact. Dr. Li will conclude by outlining a path toward developing trustworthy AI systems that can meaningfully advance patient care.