I’m Robert and I’m the CTO and Chief Architect at Castlight Health, as well as a Technical Advisor to Computable.
At Castlight, we do a lot of interesting ML work to reverse engineer the negotiated rates between health plans and health care providers so that we can give our users accurate estimates of their out-of-pocket costs for thousands of medical, behavioral health, and dental procedures, as well as for many prescription drugs. We think people should know how much these procedures and drugs are likely to cost before they are performed or prescribed. What a crazy idea, right? We also use ML extensively to classify users into segments, e.g., at-risk for diabetes, so that we can make useful recommendations to help them get better, higher-value care. Most recently, we’ve been using RNNs to predict future diagnoses and procedures. I speak about this work relatively frequently at conferences. About 17 million people (mostly in the US) have access to Castlight through their employer or health plan.
Using ML in the health care space is especially challenging due to the extremely sensitive nature of many of the data sets. Add to that the regulation that results in increased legal, security, and privacy compliance efforts. While all very important, this makes it very hard for small teams with amazing ideas to get started. I’m very interested in helping to enable access to this data in a way that preserves the privacy of individuals and allows them to maintain control of their data.