Last August, major insurer Anthem announced a 12-month trial with startup Doc.ai to see if artificial intelligence could find patterns in people’s experience of allergies. With the trial coming to its scheduled completion, Anthem describes the partnership as a “virtuous cycle” that lets the insurer benefit from data and insights it couldn’t acquire on its own.
“We are sitting on this amazing data set that we are allowed to use for doing process improvements to health care, but which will be very restricted,” said Ted Goldstein, chief artificial intelligence officer for Anthem AI, at Transform 2019.
An AI-driven startup has more leeway than a traditional medical company would. “But a startup company, of course, can give the apps out to the world and collect data … they can develop hypotheses there, which we can then look and observe to see in our data set,” he said.
Doc.ai aims to let users manage their own health data, which it says is stored locally on users’ devices to prevent hacking. The company has also enrolled interested users in studies on Crohn’s disease and colitis.
“We’re doing this because we have not yet had the opportunity to collect all that real data in one place to try to understand how [it] connects,” Doc.ai cofounder and COO Sam De Brouwer said of her company’s mission.
She said even small amounts of data can help predict a user’s future health. “It is possible already in health care, just based on your age and sex, to give you some level of predictions about health risks and other things,” she said.
And not all data has to be personal. The company plans to add location-related health factors, like the water quality where a user lives. Goldstein said the partnership relates in part to changes in the economics of health care.
“One of the things that I think the lessons of the Affordable Care Act, Obamacare, [have] had is that we are now in a position of learning that keeping people healthy is the best way to make money,” he said.
Aligning with this new paradigm requires converting to more automated processes, he added. “Anything today that requires a lot of human intervention, if it can be automated, we think that that’s a good thing.”