Chairs: Laya Rafiee Sevyeri & Steven Fernandes
Mission and objectives
AI can support diagnosis, risk prediction, clinical decision-making, and public health services. However, benchmark performance alone does not establish reliability or safety in practice. Changes in patient populations, clinical workflows, and data acquisition can affect performance, while bias, poorly calibrated predictions, and limited transparency can undermine trust.
This workshop brings together researchers, healthcare professionals, industry practitioners, and policy specialists to examine how healthcare AI can be developed, evaluated, and deployed responsibly. Its scope spans predictive models, foundation models, and generative AI using clinical records, medical images, and other health data.
- Examine methodological and practical challenges affecting robustness, trustworthiness, and safety.
- Discuss evidence for reliability, failure detection, and communication of uncertainty.
- Explore technical safeguards, human oversight, and governance for responsible use.
- Identify research priorities and opportunities for interdisciplinary collaboration.
Intended audience
AI and machine learning researchers, healthcare professionals, digital public health practitioners, industry researchers and developers, policymakers, students, and early-career researchers. Contributions from diverse healthcare settings and resource contexts are welcome.
Expected outcomes
Participants will gain a clearer understanding of how to assess healthcare AI beyond average predictive performance, recognize failures and bias, and connect technical safeguards with clinical and public health needs. The talks, panel, and poster discussions will identify open questions and opportunities for collaboration.
Format and schedule
A proposed 90-minute session with two invited talks and a moderated panel. Both invited speakers will participate in the panel. The final duration is subject to assignment by the DPH organisers.
| Duration | Activity |
|---|---|
| 5 min | Opening remarks – Laya Rafiee Sevyeri |
| 20 min | Invited talk 1 |
| 20 min | Invited talk 2 |
| 40 min | Panel discussion, including audience questions |
| 5 min | Closing remarks – Steven Fernandes |
Accepted contributions are provisionally planned for poster presentation. Poster display before the session and discussion afterward remain subject to agreement with the conference organisers. These arrangements are additional to the proposed 90-minute session.
Speakers
McGill University, Canada
Associate Professor Steven Fernandes
Creighton University, USA