Liam Kaufman is not the first CEO whose idea for a product emerged from cutting-edge university research. But his innovative technology may revolutionize the way dementia and other cognitive diseases are diagnosed and monitored.
WinterLight Labs, cofounded by Kaufman with Prof. Frank Rudzicz and Maria Yancheva & Katie Frasier, two graduate students from the University of Toronto, is developing a proprietary AI diagnostic platform that analyzes natural speech to detect and monitor dementia, aphasia and other conditions involving mental decline. All that's needed is a one-to-five-minute sample of natural speech.
Currently, patients thought to be developing dementia are evaluated using manual pen-and-paper tests that are comparatively clunky and subjective—not to mention stressful for older seniors, since they consist of memory-testing questions that are scored.
WinterLight's platform is far more user-friendly: the patient views a picture on a tablet and spends a few minutes describing it out loud. The verbal description is recorded and analyzed. WinterLight's technology characterizes the speaker's cognitive, acoustic and linguistic data, including lexical diversity, syntactic complexity, semantic content, and articulation, to assess the patient's cognitive status. Based on pilots in Canada, the U.S. and France, the test is 85% to 100% accurate.
Another advantage is that while pen-and-paper tests are generally only administered every six months, WinterLight's tests can be run several times a week, so are useful not only for diagnosing a condition initially, but for monitoring change over a variety of time periods.
Conceived and developed in Ontario
Founded in 2015, WinterLight is now comfortably situated in JLABS' 40,000-square-foot life science innovation centre in Toronto and has eight employees. It moved there in 2017 after “graduating” from the Rotman Creative Destruction Lab—a University of Toronto initiative that works with promising start-ups.
“The Creative Destruction Lab is part Dragon's Den and part Survivor,” says Kaufman. “Every eight weeks, you meet with investors and pitch your idea. If they like what you're doing, they agree to provide you with mentorship and guidance. If not, you're out. We were part of the machine learning stream there and made it all the way through the program.”
Kaufman has degrees in psychology, medical science and computer science. He had already conducted some research in neuroscience—and founded and sold an education technology start-up—when he met University of Toronto's Prof. Frank Rudzicz, an expert in computation linguistics, machine learning and artificial intelligence. Intrigued by Rudzicz's research, he joined two of Rudzicz's graduate students (Katie Fraser and Maria Yancheva, also cofounders of WinterLight) to continue exploring how technology could be used to analyze speech patterns for clues to cognitive problems. Eventually, the team was invited to participate in the Creative Destruction Lab.
The software is still being tested, but WinterLight is now working on commercializing it. Its founders hope to start selling it in the next few years—first to senior care homes and later to pharmaceutical companies. While senior homes and other care providers are interested in the technology for monitoring dementia in those already known to have it, pharmaceutical firms are interested in its significant diagnostic potential.
Benefiting from Ontario's innovation ecosystem
Kaufman can pinpoint three key milestones in the company's development that positively influenced its trajectory. One was raising $1.5 million from venture capital groups. Another was attracting enthusiastic media coverage. The third was closing a significant deal with a multinational pharmaceutical company. All three were the result of WinterLight's interaction with a variety of Ontario organizations that aim to support innovation.
Being based in Ontario has been a huge help, says Kaufman, both in terms of achieving goals and weathering setbacks. For example, WinterLight has received significant funding through grants, such as from the Ontario Centres of Excellence (OCE) and Ontario Brain Institute (OBI). The OBI also assisted with networking—making a number of critical introductions—and helped WinterLight collect high-quality data that has contributed to product development.
“JLABS has been incredible too,” says Kaufman. “They introduced us to our second investors and have provided mentorship through Johnson & Johnson.” The MaRS innovation hub has also been “amazing,” he adds, introducing WinterLight to investors and helping the fledgling firm attract media coverage.
Toronto: a magnet for talent
An upcoming and ongoing challenge will be finding the right expertise to fuel strategic growth: experts with machine learning credentials and experience are not exactly a dime a dozen. But Kaufman expects the company's location in Toronto to be an asset in that regard.
Ontario's public health system has also been an advantage because of the opportunity to collect large data sets.
“Often when you work with pharmaceutical companies, they own and guard the data,” says Kaufman. “The province obviously has to protect its data, from a privacy and security point of view—but from a health and science point of view, they can collaborate with firms like ours, and it's been extremely helpful.”
Start-ups will always face obstacles—but WinterLight's founders are growing more seasoned by the day, and believe they know how to overcome them and what to expect.
“Everything costs more and takes longer than you think it will,” is how Kaufman sums it up.