What we’ve learned so far on our journey to fix health care, and where we’re going next.
As we’ve talked about before, we started Buoy Health after seeing our own loved ones suffer the consequences of relying on the internet to make their medical decisions. But, recognizing there was a problem was easier than solving it. Over the past three years, we’ve struggled with a range of challenges: programming software to think like a doctor; building a brand that feels friendly but not boring; finding the best ways to deliver sensitive information — like telling someone they need to call a suicide hotline. After a lot of work, we are ready for Buoy to start taking better care of people.
What we've been doing for three years
At the beginning, our biggest challenge was building a program that could help people accurately and safely self-diagnose. Given the inaccuracy of symptom checkers at the time, we knew that we’d need to bring in brilliant people to give us the direction necessary to succeed where others had failed. Enter Ryan Adams and Warner Slack.
Ryan (PhD, Asst. Professor at Harvard’s School of Engineering) is a machine-learning guru who happens to specialize in healthcare. Warner (M.D., Professor at Harvard Medical School) is considered the “Father of Cybermedicine,” having done the first experiments between a human and a computer, where the computer acted as a doctor. This was in 1966. Under their guidance, Eddie Reyes, our CTO, hit the books, locking himself in our apartment for two months to read about advanced statistical algorithms. He emerged with a solution. As opposed to Web 1.0 symptom checkers that use simple, pre-determined question logic, Eddie explained that our algorithm would crunch countless statistics simultaneously, ranking diagnoses and questions in real-time, and choosing the next question based on statistical reasoning. If we could feed the program with the right statistics about symptoms, risk factors, and diseases, the program would be able to reason its way to a set of diagnoses, just like a real doctor. Eddie started building the algorithm, and I began pulling the necessary data from clinical papers.
Today, Buoy covers more than 1,600 diagnoses, knows 30,000 interview questions, and relies on data from more than 18,000 clinical papers covering 5,000,000 patients. In our first quality-control trial, where patients used the program in the waiting room before seeing their doctor, we agreed with the doctors’ diagnoses 90.9% of the time.
What we learned from our users’ feedback
The Buoy journey only starts with the algorithm. My two other co-founders, Adam Lathram and Nate Ren, took the lead on designing Buoy’s look and feel, which raised even more questions. What could Buoy do to make people feel listened to? How could a computer program convey empathy and compassion? What happened if someone misinterpreted Buoy’s questions? What would incentivize people to answer enough questions for Buoy to get a detailed understanding of the situation?
To find answers, our approach was to create mock solutions in PowerPoint and hit up coffee shops, libraries, and cafeterias to test our theories. We talked to a lot of people, and through these conversations, something started to become apparent: not knowing what was causing their symptoms was a huge source of anxiety for people. If somehow our program could show them a path back to healthiness, we’d be playing a role in caring for these people. Technology — sometimes vilified as job-stealing and, more recently, as a potential threat to humanity — could actually care for people. This revelation became the basis for our company’s mission: create technology with heart for the health of every person in the world.
Why we’re excited for people to use buoy
Things really started to pick up last summer. Our team had grown into a close-knit group of nine, each with a unique skill set but a shared drive to change health care. The product itself had overcome a number of pain points and undergone a lot of testing. We “soft launched” — by opening the product to usage without telling too many people about it — in August. And, of course, as users started coming in, we found new pain points to solve. By January of this year, three astounding things started to happen.
First, we were getting fan mail from our friends and family. One friend and his girlfriend used Buoy when they got sick in South Africa, avoiding an unnecessary hospital visit abroad. Another, an adult primary care doctor, used it to help triage his daughter. My own aunt used Buoy to understand chest pain she had been experiencing.
Second, as a 4th year medical student, I get asked medical questions pretty regularly. But the number of phone calls I was receiving from my family and friends went to nearly zero, because they were turning to Buoy instead.
Third, some people were coming to Buoy with serious, often heartbreaking, concerns. From depression to thoughts of suicide to body shaming issues, people were opening up about their biggest demons to their computer, and Buoy was there to help.
This is when we realized that Buoy was working. After three years of nonstop development, we had something that was useful to people’s health and happiness. And so we decided to start telling our story and to get Buoy into as many hands as possible.
What we're planning for the future
As proud as we are of what we’ve created and its ability to care for you, we’re even more excited to show you what’s coming next. From providing follow-up that allows you to track your symptoms over time to installing the tools that allow Buoy to learn and get better each time you use it, we are doubling down on creating the most personalized, caring health experience there is. For all of you Disney fans out there, we’re building the brain for a real-life Baymax. Our ultimate vision is to make Buoy available for every single person in the world, for free, so that we are with you, caring for you, whenever you get sick.
– Andrew, CEO