Karl Kaufmann, an emergency room physician at Washington’s Valley Medical Center, spends a lot of time at the end of a shift doing administrative work, like typing words and checking boxes into electronic medical records.

But recently, he’s been testing out a virtual medical scribe called SayKara.

SayKara was developed by a group of former employees from companies like speech recognition giant Nuance and Amazon. The team is based in Seattle and is launching this week after several years quietly developing the technology and securing a $2.5 million seed round from local investment firm Madrona Venture Group.

“I use SayKara in two ways, both to recap a patient visit and to incorporate the pertinent details of a patient interaction into the medical record,” said Kaufmann. Similarly to Amazon Alexa, SayKara starts working when a physician says a hot word, like “OK Kara” or taps on the app that runs on iOS devices, including iPhone and iPad.

Kaufmann was one of the first to try out SayKara, which aims to be an alternative to human scribes and existing dictation tools like Nuance’s Dragon. The goal is to accurately transcribe audio to text, parse the information to make it structured, and insert it cleanly into an electronic health record.

Voice is rapidly becoming big business in health care, as medical systems look for new ways to help doctors focus on the patient interaction, rather than the computer. Studies have shown that doctors today spend about a quarter of their time on the patient visit, with nearly half on desk work and charting in the electronic health record.

As a result, hospitals are already tinkering with Amazon’s voice technology Alexa. And businesses like Augmedix are leveraging Google‘s Glass hardware to offer electronic health record support with a remote scribe.

But SayKara says it’s taking a different approach. Unlike Amazon, it is narrowly focused on the health-care space. And unlike more established competitors like Augmedix, it doesn’t rely on human transcribers, which are expensive and take months to train.

SayKara CEO Harjinder Sandhu says the first inspiration for the technology was Amazon Alexa, but that the team realized early on it needed to focus on the needs of the medical sector to get it right. In order to sell to hospitals, SayKara needs to meet compliance, security and privacy requirements, and ensure that it doesn’t make seemingly small mistakes that can actually have an enormous effect (a difference is “hypo” versus “hyper”).

“The hurdles in health care are very high,” admits Sandhu.

But Sandhu believes that the current technology is good enough for physicians to use. “We’re taking the state of the art in speech recognition and machine learning and we’re applying it,” he explained. “The key is to couple it with deep knowledge of how physicians work and the information that is relevant to them.”

The company hasn’t figured out a price yet, but Sandhu maintains that it cost a fraction of a human scribe.

Ralph Pascualy, former CEO of Swedish Medical Group, is one of the first to sign up to use the tool. In his view, so much of his doctors’ time is wasted in medical records that any company that can eat “even halfway into the problem” is worth the cost. He also appreciates SayKara’s approach in not claiming to replace doctors, but to help them with specific challenges.

“Unlike some AI tools, this company isn’t attempting to boil the ocean,” he said. “I think physicians will be more comfortable with that.”

Source: Tech CNBC
This Seattle start-up is building an Amazon Alexa for health care