AI in Medicine: Highlights from Session #3
Dr. Trevor Day explored the legal and ethical challenges of integrating AI and digital health. Joining the session were experts from the Canadian Medical Protective Association (CMPA) and leading legal professionals.
Speakers:
- Dr. Christine De Maria, CMPA Physician Advisor
- Rebecca Bromwich, lawyer and legal scholar
Watch now:
Key Takeaways:
AI in Health care = Powerful but Risky
- AI can summarize charts, predict outcomes and assist with documentation but it cannot replace clinical judgment. Physicians remain fully accountable for patient care.
Legal & Privacy Realities
- AI tools introduce hidden risks: privacy breaches, bias and hallucinations. Alberta’s OIPC requires a Privacy Impact Assessment (PIA) for AI scribes under the Health Information Act - don’t confuse this with federal compliance.
Bias is Real and Dangerous
- A 2025 Nature Medicine study revealed shocking disparities: patients of colour were recommended for more invasive procedures; LGBTQIA+ patients were flagged for mental health far more often for identical symptoms.
Liability & Consent Are Non-Negotiable
- Physicians must review, verify and document AI outputs. Obtain informed consent every time AI is used. Patients are able to opt out without it affecting care.
Practical Guidance for Safe Adoption
- Start small, monitor performance and choose tools that make it easy to review and annotate notes. CMPA and CPSA have checklists and Q&A guides to help.
Emerging Legal Trends
- U.S. cases show lawsuits for both using AI incorrectly and not using AI when expected. Standards of care are evolving - stay ahead.
Communication = Risk Management
- AI scribes can free you to focus on patients, improving rapport and reducing complaints. But over-inclusive notes or unverified content can backfire - always clean up and confirm accuracy.
See other videos in this series.