AI can improve Ontario’s health care system at lower costs

Share on:

Getting more for less is the consumer Holy Grail, but it’s rarely fulfilled.

Yet by integrating artificial intelligence (AI) solutions into our health system, this is exactly what Canadians could get: better care at lower costs, freeing up resources to deliver faster and more integrated care.

However, to fully realize health AI’s potential, researchers need better access to data collected by our publicly funded health system — something that current policies make challenging.

The pandemic highlighted the dangers of a strained health system and the domino effect on services and staff. Across Canada, health spending is the single largest budget item. Every dollar not used as efficiently or effectively as possible is money taken from other programs. To address current system pressures, we need sustainable solutions that don’t pit the needs of one service against another.

Investing in health AI innovation is not a silver bullet. But it could go a long way toward alleviating volume pressures and wait times facing the Canadian health system.

Identifying illnesses earlier or predicting the risk of someone becoming ill helps improve care and better allocate resources. But we can only achieve such improvements by training our AI models with as much patient data as possible. To use an analogy, would we want doctors to have more or less training?

Ontario researchers and clinicians have taken steps toward integrating AI into existing health services. Toronto’s Vector Institute helped the University Health Network integrate AI into MEDLY, an application that remotely manages patients with congestive heart failure, generating alerts when a patient’s health is at risk, reducing hospital visits, and keeping volumes down while allowing clinicians to quickly respond to patients.

AI models improve as they learn. For example, for a model that can predict lung collapse, the greater the number of X-ray images we use to train it — and the greater the diversity of the patient data — the more accurate its diagnoses and predictions will be.

But health research has traditionally relied on “data minimization” — giving researchers the least amount of de-identified patient information needed to test a theory. This practice has been underpinned by conservative interpretations of what constitutes the minimum amount of data for research, and of existing privacy laws, most of which were drafted before modern data science techniques became widespread.

Redefining what constitutes the minimum amount of data required to suit modern data science methods would go a long way to broadening researchers’ access to data.

A 2021 Canada Health Infoway poll suggests that the majority of Canadians surveyed are comfortable with AI being used as a health tool. And the Ontario Health Data Council (OHDC)’s recent report calls for “governance and policies for health data as a public good.”

Significant progress has been made since Vector first highlighted this issue in early 2020. Partnering with the GEMINI hospital data sharing network based at Unity Health Toronto has allowed Vector researchers to develop cutting edge AI models, including studies related to COVID-19.

A curated and standardized collection of de-identified in-patient data from 30 Ontario hospitals, GEMINI is one of Canada’s largest collections of hospital data. Its continued growth will expand opportunities for projects that benefit patients and Canadian health systems, while protecting patient privacy and meeting stringent research ethics standards.

The use of AI to identify and treat COVID-19 variants during the pandemic highlighted the effects of freer sharing of data. This is an opportunity for Canada to continue to drive policies that improve patient outcomes and be a global leader in health AI research and application.

Technology moves fast. AI is evolving at a pace that policymakers haven’t been able to match. About 85 per cent of Canadians surveyed agree that investing in innovative health technologies is important.

Let’s use the lessons of the pandemic to improve access to health data for AI research and make policy changes for a better and more efficient health system for us all.