Deep Genomics is using AI to predict disease


Company’s software replicates how a cell functions and what mutations can lead to.

deep genomics

Deep Genomics:

Read in 1 minute

Brendan Frey has first-hand experience with the unpredictability and uncertainty of the medical profession. In 2002, doctors discovered a genetic mutation in his unborn baby, but they couldn’t tell him or his wife what it might lead to. They weren’t sure if it would be a problem or not.

“It was just scary,” he says. “It didn’t help us make a decision in terms of that information.”

The situation wasn’t an aberration, he adds, but rather widespread. Frey says labs generally disagree between 15 and 50 per cent of the time on whether a genetic mutation will cause issues. Stats like that don’t inspire anyone with confidence.

Frey, who earned a PhD in machine learning at the University of Toronto, had become a professor by this point. He decided to change his career direction from vision and speech processing and reoriented his research group to focus on the human genome.

More than a decade later, in 2015, the new path eventually resulted in the founding of his Toronto-based AI startup, Deep Genomics.

“I wanted to do something more transformative in terms of impacting society than detecting cats in YouTube videos or whatever it was I was working on,” he says.

Frey and his fellow researchers have developed AI that they believe can replicate how cells behave, and more importantly, predict how mutations might affect them. The software can sift through hundreds of millions of genetic mutations to find the ones that will lead to disease, which is not something a human analyst could likely do.

“It’s like taking a whole bunch of books … and trying to find a particular letter that tells you which book your child likes,” Frey says. “It just seems absurd.”

Deep Genomics has attracted $5.5 million in seed funding from U.S. venture firm True Ventures. The company is up to 20 employees and expects to double in size in the next 18 months.

The company is in the final stages of deciding which market it wants to go after, with an announcement due in the next month. Frey says Deep Genomics’ technology could be sold to a variety of buyers, from therapeutics and diagnostic providers to insurance companies, who could use it to better inform customers of the illnesses they could be susceptible to.

Frey and his colleagues have grand ambitions for their technology regardless of which direction they decide to go in.

“The goal is to revolutionize medicine,” he says. “[AI] will become one of the most valuable components of medicine.”

Tweet about this on TwitterShare on FacebookShare on LinkedIn

Leave a comment

Your email address will not be published.