AI Startup Saykara Adds Autonomy to Medical Voice Assistant
Healthcare artificial intelligence startup Saykara announced this week that its voice assistant can now automatically extract meaning from conversations between doctors and patients. This autonomy marks a significant step toward the startup’s goal of a doctor’s assistant that requires little or no human input to be helpful.
Autonomous and Ambient
Saykara describes its virtual assistant, named Kara, as a way for doctors to do away with the need to take notes during patient conversations or fill out forms about the patient. Earlier versions of the voice assistant made it ambient, able to operate without the need for human commands. The new version adds real-time autonomous analysis of meaning in the conversation through what the company calls ‘discretization.’ The interview is broken down into data points and turned into an easily understood report that can then be examined later on by the doctor.
In its ideal form, Kara would make any human note-taking or organizing unnecessary. That’s not a minor annoyance. The amount of time that doctors spend on data entry is directly tied to how quickly they burn out. These days, doctors can spend twice as much time doing administrative work as patient care, according to an Annals of Family Medicine study. Many end up leaving healthcare mainly because of the drain they feel from doing paperwork. It’s the same exhaustion that can lead to doctors making mistakes they would never make if they didn’t feel drowned in paper. The company claims using Saykara shortens the time doctors spend on paperwork by two to three hours a day while reducing the documents doctors fill out by 70%.
“Saykara has been built on the core belief that conversational AI can not only solve physician burnout and improve job satisfaction, but that it is the single most important foundational technology for healthcare in the next decade,” Saykara CEO Harjinder Sandhu said in a statement. “Up until now, AI has made an impact with access to medical imaging, lab results and patient history. Now we’re at AI’s next frontier in medicine. Much like autonomous cars learned from shadowing human drivers, Kara is shadowing doctors and learning from the incredibly rich data available from exam room conversations.”
Doctor AI
A growing number of companies are starting to offer a voice assistant for doctors to help relieve them of some of their administrative tasks. Amazon came out in December with Amazon Transcribe Medical, an automated transcription service for medical professionals, while Nuance and Microsoft recently partnered to upgrade and merge Nuance’s Dragon Medical Virtual Assistant with Microsoft’s Azure platform. Dragon combines transcription with automatically filling in patient details. Saykara also has to contend with other startups bringing voice and AI tech to doctors, such as Suki, which has raised $20 million and partnered with both Amazon and Google to extend its technology to new clients. Since Sandhu founded Saykara in 2016, the company has raised about $9 million from investors, including SpringRock Ventures, Madrona Venture Group, and New York-Presbyterian Hospital. It now counts 25 healthcare organizations among its clients.
Sandhu brings a lot of experience in this space to Saykara, however, particularly when it comes to Nuance. He previously founded medical transcription AI startup MedRemote in 2000 before selling it to Nuance in 2005 and spending the five years as Nuance’s chief technologist for healthcare R&D. Saykara claims that none of its potential rivals combine the autonomy and ambient nature of its voice assistant. Their voice technology requires dictation to create a transcript or commands to parse what was said in a conversation, instead of it all happening automatically and in real time. Saykara still has to compete with more familiar names though, and the kind of bundled technology that only the more prominent companies can offer.
Follow @voicebotai Follow @erichschwartz
Amazon Launches Medical Transcription Service in Direct Competition with Nuance
