How AI is helping treat people with depression

This week a Canadian startup, aifred health, won one of two Milestone Awards in the first year of the four year long IBM Watson AI XPrize. The XPrize is a competition to create AI systems to solve some of the world's biggest challenges. The Montreal-based company, who is developing a system to help doctors make treatment plans for patients with depression, will continue on to the next year of the competition as a top 10 finalist.

Sonia Israel is the Director of Partnerships and one of the co-founders of aifred health, as well as a neuroscience student at McGill University and the Douglas Mental Health University Institute. Sonia says that treating depression is one of, as the XPrize website describes it, "the world's grand challenges," because of the debilitating effect it has on so many. "Depression is actually the leading cause of disability. 11 per cent of the world's population, at some point in their lives, will suffer with an episode of depression."

And while it affects so many, treatments for depression can vary hugely from person to person. "It's essentially trial and error," Sonia says. "It's an educated 'guess and check' approach that doctors are using with their patients right now for prescribing mental health treatments, especially medications.

"Regular antidepressants normally take about three to four weeks to start taking effect. So the doctor has to prescribe it, and then you go home, you wait, and if you have side effects you go back, or you go back, it doesn't work, they say ok, let's try something else."

aifred health's system uses deep learning — the same technology that powers Google's Go playing program AlphaGo — to make suggestions based on medical research and the patient's medical history. "What we're trying to do," Sonia says, "is provide doctors with an evidence-based tool, a software tool, that they can use to match the best mental health treatment to each and every one of their individual patients, using that patient's physiological health profile."

While a full version of the tool is still a while away, aifred health plans to release a light version of their tool sometime next year. In the meantime, they are also developing clinical trials to test their tool's effectiveness.

Currently, aifred health is focused on depression treatment, but they don't believe their tool will be limited to just that particular disorder. "We want to expand to different areas in psychiatry," Sonia says. "We're thinking psychosis next. And eventually we would be interested in fields outside of psychiatry."