Canada’s AI strategy aims to make health data more accessible. How will it protect privacy?

The federal government is promising to invest $100 million to make Canadians’ health information — including imaging, medication records and vitals — more accessible to researchers as part of its national AI strategy. 

Connecting data across the country will help researchers conduct clinical trials, test AI health tools and drive innovation in treatment and diagnosis, federal Minister of Artificial Intelligence Evan Solomon said in a statement to CBC News. 

But the initiative will also have to balance privacy considerations. Experts say this will involve removing identifying information, giving access to only trusted researchers and investing in Canadian-owned platforms and data centres. 

To be used for research, the data “needs to be de-identified or anonymized, to protect patient privacy, and governed carefully,” said Dr. Amol Verma, a physician and co-leader of Vital, the platform that is receiving federal funding.

In his statement to CBC, Solomon noted “trust has to be earned by being clear about both the public benefit and the safeguards,” and said the data will be kept in secure infrastructure for only approved researchers to access.

Why is the government doing this? 

The funding announced on Tuesday will expand Vital, which currently connects data from electronic health records in Ontario, Alberta and Quebec to make it more accessible to researchers.

“Canada’s health-care system does not use data well to improve itself or to advance innovation and discovery,” Verma said. Information is stored in separate hospitals and clinics, making it hard to access and study.

The country’s health data is incredibly valuable, because Canada’s population is diverse and because our universal health-care system means data from a large number of patients is collected, according to Khaled El Emam, a University of Ottawa professor who runs the Ottawa Medical AI Research Institute. 

“We have an advantage and we need to take advantage of that for improving our health,” he said.

WATCH | AI could improve patient experience in Canada:

Could AI help manage overcrowding in N.B. hospitals?

Researchers at the Université de Moncton believe artificial intelligence is the answer to improve the way patients are processed in and out of hospitals, which could reduce overcrowding in acute care beds.

If researchers have more access to health data, they can use that information to detect outbreaks, identify new treatments and develop new drugs — and AI can boost those discoveries, he said.

More accessible data will also attract studies from around the world, meaning Canada could get first access to new treatments and drugs, and could tap into a clinical trials market worth billions of dollars, according to Dr. Fahad Razak, also a physician and Vital co-leader.

Concerns about access, transparency

Canadians’ trust in AI in the health-care space “exists on a spectrum,” cautioned Cristyana Aloysious, co-chairperson of the steering committee for OurHealthData, an organization pushing for people-centred health data governance. 

Concerns typically include worries about who has access to the data, why it’s being used and transparency if there is a privacy breach, she said. 

“They want confidence that if something goes wrong … there’s clear accountability.” 

Another concern among researchers is that algorithms could analyze even anonymous data in a way that is biased against people with shared characteristics, according to Dr. Sheryl Spithoff, a physician and assistant professor in the department of family and community medicine at the University of Toronto.

For example, a 2022 review out of Brown University in the U.S. highlighted how AI could fail to adequately diagnose skin diseases for Black and brown patients, or misidentify hot spots of communicable diseases because of racial bias in an algorithm.

As well, Spithoff said, there’s always a risk of “re-identification,” especially if computers become powerful enough to undo the anonymization process and expose identifying information about people.

Razak noted that a lot of Canadian data is stored on American platforms and servers, and American technology companies can be legally forced to open up their data to the U.S. government.

In theory, that could lead to Canadians’ health data being handed over to American officials, like immigration and customs enforcement, he said.

Privacy protections work ‘very well’ 

To address concerns about foreign access to sensitive information, the federal government is investing in Vital and other Canadian-led health data platforms and infrastructure, Solomon said. 

In the meantime, Verma said protections include storing all the data in Canada, encrypting it and following privacy best practices. 

Key identifiers — like names, birth dates and health insurance numbers — won’t be included in the datasets researchers can access, he said.

The initiative will also ensure strong cybersecurity around the centres where data is stored, and access to the information will only be given to trusted researchers from Canadian universities or research institutions, Verma said.

Researchers will access the data by logging into a web portal that’s separate from the rest of the internet, he added, and studies will be reviewed again after publication to ensure they don’t include sensitive information. 

The privacy risks of the initiative are small, El Emam said.

“It’s a process that has been applied for a long time,” he said. “It’s worked very well in practice.”