Commentary
Dr. Jehangir Appoo

One of the main problems with Canada’s healthcare system is the dearth of family doctors. The growing popularity of artificial intelligence raises an interesting question—can AI help solve Canada’s physician shortage?

According to a 2023 Unity Health Toronto survey, approximately 20 percent of Canadians do not have a regular family doctor or nurse practitioner compared to less than 5 percent of residents in Denmark, the Netherlands and Norway. And only 46 percent of Canadians report being able to make an appointment on the same day they are sick—ranking Canada 9th out of 10 OECD countries with universal health care.

Delayed access to care in Canada is not only due to a scarcity of family doctors but also due to the type of work-family doctors chooses to pursue. Many family doctors who train as primary care physicians choose to practise in non-primary care medical occupations such as hospitalist, surgical assistant and clinical associate with specialty practices, which suggests that family doctors find other aspects of the work more interesting than primary care. Young people generally want to enter careers where they can do creative work on important problems and work with other motivated, talented people. Subsequently, the best way to improve access to family doctors may be to make the profession more attractive to top talent.

So, can AI reduce the burden of mundane tasks and thereby allow family doctors to do more creative work—which in turn will increase the number of family doctors in Canada?

At it’s core, AI is a tool for prediction.

And a significant portion of medical work involves making diagnoses or “predicting” what is wrong. While patients sometimes use a form of AI (known as Large Language Model or LLM technology) to self-diagnose, some physicians globally are using LLMs to generate a list of diagnoses to ensure they don’t miss something potentially important.

Indeed, a personalized LLM, which knows the individual patient’s history, could triage problems more efficiently. As a result, family doctors could transition to the creative and rewarding work of creating a meaningful treatment plan that helps patients. As we approach the two-year anniversary of ChatGPT’s release, the technology today remains in its infancy and will likely improve in the same way the Internet has grown more useful over time.

But a word of caution. While a new paradigm of care augmented by AI may reinvent the work of family doctors to once again make it an attractive specialty for top talent, AI won’t alleviate the current crisis tomorrow. The regulatory process in medicine takes time.

However, government policy can play a key role in the adoption of LLM AI in Canada in a few different ways. Government can financially support training programs for family doctors to help integrate LLM AI into Canada’s individual electronic medical records (EMR). Government can also foster the interoperability of existing provincial EMR systems so they more seamlessly exchange patient information. AI tools within a given EMR may then leverage data across many EMRs to help create powerful, population-specific prediction tools that may increase the quality of care, shorten patient appointments, streamline medical records, and thereby improve patient access to care.

We’re only beginning to understand the potential of AI to improve health care in Canada and around the world. And while AI may help attract top talent to primary care by making the work more interesting, governments in Canada must continue to work on more conventional strategies to improve health-care access while supporting the careful incorporation of AI models including LLMs into medical care.

Dr. Jehangir Appoo is an Associate Research Professor at the University of Calgary.