HIROC Conference 2025: AI in Healthcare

This is the third and final article of our recap series on the HIROC Conference.
If you missed the first two, be sure to check out part one and part two!
We now delve into one of the hottest topics today—AI and the key considerations for adopting it at your healthcare organization, including liability, rules and regulations, and risk management.
Melanie de Wit, Chief Legal Officer, Unity Health Toronto, began by describing a few of the over 50 AI tools Unity Health uses, including ones that detect patient deterioration and help with notetaking. All of these AI tools presented solutions to challenges identified by their frontline clinicians.
As de Wit explained, “We choose problems where AI can function to either improve patient outcomes, reduce costs, improve flow, or optimize the patient or healthcare provider experience. So, for us, the benefits of AI are clear, but I think as a sector and as individual organizations, we have to have a high degree of confidence that those benefits are going to reliably exceed the risks of clinical AI.”

Next, a panel discussion featured speakers de Wit; Mary Jane Dykeman, Managing Partner, INQ Law; George Wray, Partner, Borden Ladner Gervais; and moderator Leo Dias, HIROC’s VP, Performance Excellence and Information Services.
Key Takeaways from the Expert Panel
Internal governance and oversight mechanisms should be in place before using AI tools in clinical practice. Creating a board subcommittee on AI and an AI governance committee could be a good place to start, as well as setting a risk tolerance for each risk category.
This way teams can innovate with fewer barriers while still managing the risks. Teams would only need to get approval from the AI governance committee, that reports to the board, if they hit or exceed the risk tolerance.
When making contracts with AI vendors, it is important to consider clauses on liability and performance guarantees to protect your organization. Secondary uses of data is another important clause—ensure your data is only used to train your AI model and nothing else. “It's so important to understand where your data will be and what will happen if something should go wrong,” Dykeman emphasized.
You can also manage your risk by including these in vendor contracts:
- Performance assessments
- Software update and maintenance management
- Indemnity provisions, including insurance indemnity
- Compliance with Medical Devices Regulations
- Ownership of intellectual property and copyright
Finally, the panelists shared some useful advice for adopting AI responsibly at your organization. Develop a process for how AI is approved and provide staff education and rules for using it. Start by adopting AI tools that can help solve problems identified by frontline staff; this will offer the greatest benefit and gain staff trust. Be transparent with patients, staff, and the community about how you’re using AI. And partner with other healthcare organizations that can help.
Wray added, “Canada is fundamentally a leader in health information. The unique public structure of Canada's healthcare and the delivery of it provide us with a really good, solid foundation to lead the way in healthcare and the deployment of AI.”

Thank you to all our speakers who shared their stories and expertise during our HIROC Conference and to everyone else who attended! It was great to see many of you in person for a day of knowledge-sharing as we turn the corner on patient safety, and we hope to see you again next year.
If you missed parts one and two of this recap series, read them below:
HIROC Conference 2025: Improving Resilience with Board Leadership and Cyber Security
HIROC Conference 2025: Safety Grants Supporting Innovation
By Gillian Brandon-Hart, Communications and Marketing Specialist, HIROC