AWS announces generative A.I. tool to save doctors time on paperwork
Amazon Web Services on Wednesday announced a new service for health-care software providers called AWS HealthScribe.
Attendees walk through an expo hall during Amazon Web Services' Reinvent conference at the Venetian in Las Vegas on Nov. 29, 2022.
Noah Berger | Getty Images Entertainment | Getty Images
Amazon Web Services on Wednesday announced a new service for health-care software providers called AWS HealthScribe, which uses generative artificial intelligence and speech recognition to automatically draft clinical documentation.
The service aims to save health-care workers time using AI-generated transcripts and summaries of patient visits, which can then be entered into the electronic health record system. AWS HealthScribe can also extract notable medical terms, medications and other key details, according to the company, and physicians can double-check each line of generated text with the original transcript.
Clinical documentation is a major pain point for doctors and nurses. A study funded by the American Medical Association in 2016 found that for every hour a physician spent with a patient, they spent an additional two hours on administrative work. The study said physicians also tend to spend an additional one to two hours doing clerical work outside of working hours, which many in the industry refer to as "pajama time."
As a result, several companies like Microsoft's Nuance Communications, and now AWS, have been working to build solutions to reduce this administrative burden.
"It is clear that generative AI has the power to transform the health-care and life sciences industry in many ways," Swami Sivasubramanian, AWS's vice president of database, analytics and machine-learning services said at during a keynote speech at AWS Summit New York Wednesday.
Microsoft's Nuance announced its generative clinical notes application, DAX Express, in March. Similar to AWS HealthScribe, Dax Express automatically generates a draft of a clinical note within seconds after a patient visit. It can record a conversation between a doctor and a patient in real time and create a note using a combination of existing AI and OpenAI's newest model, GPT-4.
With both services, physicians can review the AI-generated notes before entering them into the electronic health record system.
AWS HealthScribe is powered by Amazon Bedrock, which is the company's service for building generative AI applications. AWS said Wednesday that AWS HealthScribe is compliant with HIPAA and does not retain any customer information. Customers can also choose where they would like to store their clinical documentation.
The cost of the service will vary, as AWS HealthScribe is available on a pay-as-you-go basis, according to a company blog post. Customers will be charged based on the seconds of audio processed per month.
Several organizations are already using HealthScribe, according to AWS, including the software company 3M Health Information Systems. Detlef Koll, 3M's vice president of global R&D, said the company has been working with AWS since this past fall to introduce the technology responsibly, ideally without causing disruptions in documentation quality.
Koll said the technology is "excellent," in his experience, and that the tool will not function as decision support or change the care that patients receive.
"Technology is an enabler for a solution, it's not the solution," Koll told CNBC in an interview Wednesday.
The initial use cases for AWS HealthScribe were geared toward general medicine and orthopedics specialties, according to an AWS company release. The technology is available in a limited private preview capacity starting Wednesday, and Tehsin Syed, general manager of health AI at AWS, said the company plans to work closely with its customers to determine plans for expanding access.
"We update the underlying technology based on the feedback that we're getting," Syed told CNBC. He added, "From an adoption perspective, I think there's a lot of interest, and we want to be very careful about making sure it's going to work at scale in the right way."