You've touched upon a significant issue. Right now, as it stands, we have AI being deployed across the country, but in a very patchy distribution. There's very little consistency in how and where it's going and how the AI tool is monitored once it's been deployed.
The ones being deployed currently are for things like picking up bleeds in the brain. For someone coming into the emergency department, the scans will be run through an algorithm that can pick up brain bleeds. You have others assisting in the lung cancer screening program to pick up nodules. Then there are AI tools helping to identify abnormalities in mammography. Those are the ones that are diagnostic.
Then there are others in the background that may sort the imaging studies for prioritization or help with the actual booking and scheduling of patients. There are multiple levels.
The Canadian Association of Radiologists has proposed the Health AI Validation Network. This is something we've been working on with our partners in other specialties for a number of years.
We feel that there should be oversight. Products are regulated by Health Canada and they get approval, but there's no post-market surveillance of what happens to them when they go into the market. In addition to assisting the health care system and making it more efficient, there does need to be oversight.
