As a kid, I assumed that doctors diagnosed patients by asking a series of questions, then following the answers down a decision tree until they arrived at the right diagnosis. I now realize clinical recommendations are not that simple, but I believe statistical modeling on the large amounts of historical data could still help give better recommendations. One idea would be to offer a website where patients answer 20 questions about their ailment, then receive a best-guess diagnosis. This would save lots of time and money over going to a clinic and taking the time from a busy (and expensive) doctor.

I also see this as a chance to incorporate cost considerations into clinical recommendations. Right now, when I go to a doctor, I can’t see the true price of possible treatments. This lack of price transparency often allows patients to overspend on healthcare, driving up overall healthcare costs. We could use the vast amounts of data from large insurance companies to help patients make more cost-conscious treatment decisions, lowering costs improving care.