- Amazon Web Services unveiled a medical speech recognition service that converts doctor-patient conversations to text in a bid to streamline onerous clinical documentation and lower administrative costs.
- Called Amazon Transcribe Medical, the product is HIPAA-eligible and integrates into voice-enabled applications and microphones via an application programming interface. It was developed in partnership with AWS clients, including EHR giant Cerner and voice-enabled clinical assistant company Suki, and announced Monday during AWS' annual conference in Las Vegas.
- AWS' seven HIPAA-eligible offerings already included Amazon Transcribe, a speech-to-text service, but Transcribe Medical has a higher level of accuracy in recognizing healthcare-specific words and medical terminology, according to the company.
Amazon is among the big tech names working to dominate the lucrative voice-to-text market, including Google and Microsoft, two cloud giants that also operate speech assistants taking notes in real time. Microsoft is developing ambient sensing technology with clinical documentation company Nuance, focusing on how to automatically input health data into the correct spot in the EHR.
Google has been working with Stanford University for two years on transcription of medical conversations. Additionally Suki, which worked with AWS to develop Transcribe Medical, has an ongoing partnership with Google Cloud, though CEO Punit Soni told Healthcare Dive in October it's not an exclusive relationship.
Amazon and other non-traditional entrants are looking to diversify revenue streams to bolster their bottom lines. Amazon saw its profit fall year over year in the third quarter for the first time since 2017, but AWS continued to be a major profit driver with operating income of $2.26 billion in the quarter, almost 72% of Amazon's total operating profit.
The problem of administration overload is well known: Clinicians can spend up to an average of six hours per day writing notes for EHR data entry alone. Several startups and established players, including Amazon, are working on ways to ameliorate the documentation workflow, which is full of repetitive typing and routine actions that can be automated.
Unlike some other early end-to-end offerings, Transcribe Medical does not automatically insert relevant personal health information into a patient's EHR. However, according to AWS, the new service still saves time in medical data entry and can be used in tandem with Amazon Comprehend Medical, the e-commerce giant's text analytics software, to find patterns in medical text to suggest diagnoses, treatments and more.
The software isn't just targeted to providers. Cerner uses the digital voice scribe to transcribe clinician-patient interactions, then ties in its own artificial intelligence software to enter data into the correct slot in its EHR. And biotech giant Amgen has already used Transcribe Medical to review recorded calls from patients and providers to identify potential side effects of its drugs.
Amazon linked with Cerner in July to develop artificial intelligence and machine learning products in an expansion of an existing partnership. The two hinted Amazon's voice recognition and natural language processing capabilities would impact Cerner's tech offerings at the time.
AWS does not collect or use patient's health information for research and development under the terms of its business associate agreements with health customers, a fact the company explicitly laid out amid an unfriendly environment concerned about big tech's involvement in healthcare and its privacy implications.