Amazon Web Services, an Amazon subsidiary, is launching a generative artificial intelligence tool focused on clinical documentation.
The product, AWS HealthScribe, will allow healthcare software providers to more easily build applications that use speech recognition and generative AI, the company said at its AWS Summit in New York on Wednesday.
HealthScribe will enable software vendors to use a single API to create transcripts and summaries of doctor-patient discussions—along with extracting key medical details such as terms and drug information—for entry into the electronic health record for general medicine and orthopedics, AWS said. The company plans to review input from its customers and expand the tool’s application to other specialties as necessary.
AWS Vice President of Database, Analytics and Machine Learning Swami Sivasubramanian said during a keynote speech Wednesday that clinicians can validate the accuracy of the summary before entering it in the EHR. Sivasubramanian said the product is in preview mode, and the company has not yet announced when it will be generally available.
"We hear that one of the most common healthcare industry pain points is the amount of time it takes for clinicians to write detailed documentation for each patient visit," Sivasubramanian said. "We wanted to make this easier to address this pain point for clinicians."
Tehsin Syed, general manager of health AI at AWS, said in an interview following the keynote that the tool could also increase accuracy in clinical documentation by letting clinicians rely less on memory and handwritten notes.
The HIPAA-eligible technology can identify multiple speakers and segment the conversation into sections based on the chief complaint, history of present illness, assessment and plan for the condition, Syed said. The HIPAA-eligible technology can also allow clinicians to trace the origin of any generated text, according to AWS.
The tool is not foolproof, Syed said, adding that he doesn’t foresee a time when the clinician won’t need to review the AI-generated notes.
“It’s really important for the physician to have input in terms of: ‘Is this the right quality documentation?’ Even if we were 100% accurate, they might still not agree with the conclusion of what it said, or they could change their mind. I think it’s always super important to have the clinician have the final say,” Syed said.
AWS said it had secured partnerships with 3M Health Information Systems to grow 3M's ambient clinical documentation and virtual assistant software.
AWS also said it was working with ScribeEMR, a healthcare documentation company, and Babylon, an AI-enabled virtual diagnosis and medical appointments company that announced plans to go private in May, on similar offerings.
HealthScribe will be powered by Amazon Bedrock, a service for building generative AI applications. The company said its product will allow developers to integrate AI more easily in vendor offerings without needing to manage the underlying machine learning infrastructure or train against an organization’s own healthcare-specific large language model.
AWS is assessing additional ways to integrate generative AI in healthcare, but it has not released specific details. Syed said the company is seeking to balance safety and regulation with its advancements in the space.
“If you look at the Quadruple Aim–improved patient care, lower costs, improved patient experience and improved clinician experience–we’ll see a lot of application in the improved patient and clinician experience areas, because what we’re talking about are natural language interface capabilities where patients or clinicians can get time back,” Syed said.
The AWS announcement is the latest in a series of AI-related industry movement in the past week. Hippocratic AI, a generative AI company setting out to build a large language model for healthcare, announced it secured 10 partners and $15 million in new funding Tuesday.
GenHealth.ai, a generative AI company that spun off from 1upHealth in April, announced last week it secured early funding to produce its own large medical model. Venture capital firms Craft and Obvious Ventures led the round, which generated an undisclosed amount. GenHealth.ai touts its AI utility as a tool health insurance companies and providers can use for risk adjustment, care management and financial bookmarking.