Today’s guest post comes from Dr. Vivienne Ming, a theoretical neuroscientist, technologist and entrepreneur. Named one of “10 Women to Watch in Tech” in 2013 by Inc. Magazine, Dr. Ming also speaks frequently on issues of LGBT inclusion and gender in technology. She is presenting at Lilly headquarters on September 14 in Indianapolis.
A late-night dinner in New York with an aspiring entrepreneur talking through her startup idea was coming to an end. She was about to leave when she turned to say to me, “I’d drop it all if you could come up with a treatment for bipolar disorder.” I never learned why manic depression was so meaningful to her, but that night I began looking at research into bipolar. My personal purpose in life is to maximize human potential and the theoretical neuroscientist in me was intrigued by the idea of leveraging machine learning (AI) to predict manic and depressive episodes based on simple everyday data. I found a complex story — but one with hints that aggressive, continuous diagnostic testing (read: completely unrealistic testing) could offer predictive signals weeks before an initial episode.
That would have been just a side note in my many projects if not for another startup pitch I heard the next day. A small group of founders in New York has big ambitions to own the idea of emotional computing. That team is building a platform for mobile phones that can take seemingly unrelated data — gyroscope, ambient light, accelerometer, GPS, weather and so on — and estimate your current emotional state. Users don’t have to do anything; it doesn’t read your text messages or listen to your calls. You just keep your phone with you, and that platform can pick up trends. It can even map emotional states across the city.
I was intrigued and eventually joined that company’s board as Chief Science Advisor. But I had one ask: Let me use your data and technology to help people with bipolar disorder. If we cannot drag people into a lab every week for a battery of tests, what can we do with just the phone 24/7?
It turns out we can see clear signals of oncoming manic episode up to three weeks in advance of subjective report for the sufferer. I’m aiming for four weeks. The signals from the platform may be quite messy, but when they track you continuously, a rich story emerges. And it’s a story we can read at massive scale across thousands and millions of people.
Here’s a huge, unanswered question: Can knowing actually make a difference? What might be possible if we can send a message, to both the user and a confidant, that there’s a “high probability of a manic episode in the next 4 weeks”? Can they reduce medication, and the notoriously hated side effects, upping dosage only during high-risk periods? Can a check-in with a doctor, time off from work and reducing stressors and triggers actually help reduce acute incidents, job loss, divorce, institutionalization or suicide?
Until now, this has simply been a personal project. My work has been to show that the predictions are possible. The next step is to change people's lives.