Buoy Health, makers of a machine learning-powered next generation symptom checker, has raised $6.7 million in a round co-led by F Prime (the venture arm of Fidelity) and an undisclosed strategic investor in the payer space. FundRx and various angel investors also contributed.

"The unfortunate thing about trying to use Google [for symptoms] is the information you find is the information you find is either accurate or convincing, but never both," CEO Dr. Andrew Le told MobiHealthNews. "The accurate stuff like WebMD is so generic that it doesn’t do anything. Then people turn to more specific searches, which leads them to sites like Yahoo Answers. ... Those forums are so specific that if you find something that sounds like you, it’s really convincing — but probably not accurate. This leads you to get into this downward spiral of cyberchondria and uncertainty about what to do."

Le says the problem with the existing first generation of symptom checkers, like the recently sold WebMD, is that they rely on a simple decision tree ("like a really bad Choose Your Own Adventure book"). 

"If we could have replaced doctors with simple decision trees we would have done that hundreds of years ago," Le said. "But the way that a doctor thinks and reasons is much more complex than a simple tree format and we’re trying to replicate it."

Buoy is a more dynamic, interactive engine built from a data set of 18,000 clinical encounters. 

"What our program does is we’ve built an immense granular database of symptoms, risk factors, and diagnoses from over 18,000 clinical cases we read by hand, which looked at 5 million patients and the way they presented, with diagnoses," Le said. "What that allowed us to do is, after you put in your first symptom, is say which question of the 30,000 possible questions in our database has the highest statistical influence on what’s going on with you. It presents that question, you answer it, and it does the same thing in realtime again. So after a few minutes we get a pretty good understanding of what’s going on with you and we present to you three diagnoses at maximum and what to do about them."

The system rolled out in March and has picked up good traction, with about 8,000 visitors a day, 300,000 visitors total, and a 38 percent week-over-week growth. The company has also seen good data -- a 500-person clinical trial found that Buoy was able to match the diagnosis of urgent care patients to their doctor's diagnosis 90.9 percent of the time, and the company has conducted head-to-head studies with other symptom checkers that show Buoy competing favorably as well.

The next step for Buoy is to partner with payers and providers (including both traditional health systems and newer provider models like telemedicine companies and walk-in clinics) to be able to send users to a particular avenue of care based on their interaction with Buoy. The company makes money by whitelabeling the product to sit in payer and provider portals and websites, but the vision of the company is to keep building out the product until it is everpresent through a patients' healthcare journey.

"With this money, we’re helping build the tools and the pipes to build that next step seamlessly and easily," Le said. "So if you’re sick and you’re saying 'I need to go to urgent care, or to telemedicine, or to my primary care doctor', the next logical step would be to connect with providers who can provide services if you are indeed needing them. Part of this funding will be geared toward building the pipes for patients to take that next step, by picking partners in specific cities and across America and helping those people take the next step in their journey."