During a standard doctor’s appointment, a patient shares details of a condition or symptoms as the practitioner types up notes on a nearby computer. But multitasking is hard, and the patient experience may suffer as the physician balances dictation and listening.
Pedro Teixeira and Ravi Atreya experienced this firsthand as medical students in Vanderbilt’s class of 2009. Teixeira shares that when using the school’s electronic health record system, it felt cumbersome to enter the data while interacting with patients. After Teixeira and Atreya completed their Ph.Ds, joined by Michael Poku, they decided to take a leave of absence with the support of the school’s leadership.
“We just thought, we can go do residency and spend all those hours every single day doing this documentation or we can go and try to build something where it’ll be easy to get the information in, and then you could use the information for great research to ensure that patients get the best care,” says Teixeira, who is now CEO of PredictionHealth.
The startup’s AI-assisted scribing software helps doctors remain more engaged during consults as it listens and gathers comprehensive documentation and data in real-time, freeing up the doctor from typing up notes. “The format is very clean and the doctor just has to review and sign off on the note. We’re really solving both sides of that problem that we’ve unfortunately, or fortunately, had to experience ourselves for so long,” says Teixeira.
Teixeira shares that most patients don’t realize that while writing those notes, doctors often must summarize the whole conversation about symptoms, describe everything they saw during the physical exam, postulate theories of what the condition could be, and propose possible treatments. However, it’s not a complete transcript.
With their cloud-based platform, Prediction Health captures all the data during the consult, creates a more complete summarized transcript using machine learning, with room for the doctor to add any thoughts that weren’t verbalized, and includes reminders for follow-up appointments.
“We bring structure to the data. For example, if they say a specific disease, the system will know the ID number for that disease. That way if the doctor wants to ask about all the patients that have diabetes , you can do those kinds of queries. It’s not just like a note on our site, we have a lot of structured representations for that information,” says Teixeira. “We want to automate more and more of that curation process and help with quality, value-based care.”
The software can also provide a risk prediction score for patients with chronic conditions, helping doctors re-allocate their time more efficiently and providing resources for high-risk patients. “PredictionHealth can predict if a diabetic patient may have a higher blood glucose level and suggest to spend a little more time or suggest some other resources. Right now all you have is a doctor’s gut feel, but they’re also rushed from patient to patient so you don’t always know where to put those resources.”
The platform is easily accessible from a web browser and the team has plans to release a mobile version soon.
While they’re working with some health systems, the primary customer is the individual doctor, shares Teixeira, with many even paying out of pocket for the SaaS-based software. Their clients range from mid-size group medical practices to larger systems.
The Nashville-based startup has remained bootstrapped with the help of micro-grants and most recently, a $25,000 prize from this year’s 36|86 Conference student pitch competition. However, they’re preparing to go into fundraising mode in the next few months to raise a seed round.
“Our long-term vision is we believe that doctors and artificial intelligence together can do a much better job and that there’s just this missing piece, this infrastructure where you need the data in a certain way, in a certain time to be able to pull it off,” he says. “We’re very excited about trying to address that because we think that’s where the future of medicine goes.”