Health Minister Mark Holland says he’s optimistic about the role artificial intelligence can play in Canada’s health system, but says much of its promise will take time to realize.
Holland says he expects AI tools can help ease doctor workload and integrate health records from disparate providers, but that means greater digitization when many doctors still don’t send electronic scripts to pharmacies.
In a wide-ranging interview conducted Dec. 12 — before Chrystia Freeland’s bombshell resignation as finance minister and an expected cabinet shuffle Friday — Holland looked back on health headlines that dominated 2024 and what lies ahead.
He weighed in on topics including medical assistance in dying and food safety. Among his impressions: Australia’s plan to ban social media for children would not work here, and he expects Canada can find common ground with Robert F. Kennedy Jr., a vaccine skeptic favoured by U.S. president-elect Donald Trump to become his health secretary. Kennedy Jr. also says he plans to put an end to water fluoridation in the United States.
The Canadian Press: I want to ask you about Australia’s decision to ban social media for kids until they turn 16. Would you be open to exploring something similar in Canada?
Holland: I don’t think it’ll work…. The most savvy person at navigating what is and what isn’t true on social media is my 13-year-old stepchild because they have lived in this environment their entire life. And you can ban it on their phone but it’s all around. There’s computers everywhere, there’s access to this everywhere. It won’t fix it…. It’s very dubious to me that that would be successful. I understand its intent, and I share their fear, but I’m not sure that that’s the policy resolution.
CP: Health Canada says MAID applicants whose death was not reasonably foreseeable, referred to as Track 2, were more likely to live in neighbourhoods with more renters, which correlated with the Ontario chief coroner’s findings earlier in the year that these recipients had the greatest residential instability. Are you concerned about this?
Holland: The Track 2 cases are people with all kinds of very complex, severe comorbidities…. Somebody who has very complicated comorbidities and is very, very ill is probably not in a strong economic position. There’s somebody who has got expensive bills and is income insecure and (may find it) difficult to take care of themselves and (be) very reliant on others…. It could well be that it is the person’s very, very poor state of health that puts them in that situation, rather than being economically indicative that somebody who’s in a lower economic, more vulnerable situation is therefore more likely to make the choice.
CP: When it comes to AI, what would you feel comfortable seeing in a health-care setting, and what would you not feel comfortable with?
Holland: You see a walk-in clinic for one problem. You go see your family doctor for another. You go see your dental hygienist for another problem. You go see your pharmacist for something else. When you have an interconnected data system, you can run AI against those visits and look for what each of them have in common with each other.… We can run AI against it to go: OK, for this kind of condition, this kind of disease, here’s what early indicators look like. So we can catch it early.
It’s not going to replace the ability, the need, to have a physical person.
CP: Who would have access to that data? Would that mean that my pharmacist can see my data from another practitioner?
Holland: So there’s a way to both keep the data private — and you use AI to look for patterns in data that is anonymized — but then be able to, using encryption, pop it out to an individual to say you should get screened.
CP: Would you consider developing laws that specifically address AI in health care, so that doctors have something to work off of? Would you release a federal guidance document?
Holland: When you’re sharing a page in a fax machine, you’re sharing patient information, there’s all kinds of patient information that’s already being shared and already has regulations and controls around it. So there’s nothing that is different about that.
CP: But when AI involves a third-party organization — for example, a doctor paying for an AI scribe to transcribe patient encounters — the fear is that information can be moved to different jurisdictions. Some physicians just don’t really know what’s safe and what’s not.
Holland: It’s going to take notes but you’ve got to go back and verify those notes are accurate and take a quick look at them and run it through. So at this stage, that’s a tool that you have to verify and you can use to kind of accelerate things but you’re going to have to edit it. You’re going to have to make sure that it’s accurate. You can’t just rely on it. And I would argue, we’re probably a long way from the point where you can use something like AI scribe, press a button and, you know, not think about it.
CP: Customers complained about products coming from the facility at the heart of this year’s plant-based milk listeria outbreak for unrelated reasons, including possible mould, in 2018, 2019, 2023-2024. Do you feel that the Canadian Food Inspection Agency applied enough scrutiny to the facility?
Holland: There was no reason that the CFIA had to believe that listeria could be present at the facility, and therefore its risk level was very low in that circumstance. Clearly, the science has moved. We now understand that listeria could be in places that we didn’t believe it could be. So that is going to change how we inspect and is going to change the risk profile of facilities that are making similar products.
CP: South of the border, Trump has named Robert F. Kennedy Jr. as his pick for the health file. How would you approach working with him if he joins the U.S. administration?
Holland: Let’s start on what we agree with and work backwards, and let’s always keep conversations rooted in science, data and evidence…. I don’t prejudge a relationship before it’s started. I understand that there’s a lot of speculation, but let’s let the facts speak for themselves as the conversations unfold.
Source link