During its annual AWS Summit conference in New York, Amazon introduced AWS HealthScribe, an API that creates transcripts, extracts details, and generates summaries from doctor-patient discussions. These transcripts can be directly entered into an electronic health record (EHR) system. Amazon states that HealthScribe's machine learning models can convert the transcripts into patient notes, which can then be analyzed for valuable insights. The aim is to alleviate the time-consuming burden of documentation for healthcare professionals. HealthScribe identifies speaker roles, categorizes transcripts based on clinical relevance, and uses natural language processing to extract structured medical terms. Moreover, Amazon assures that HealthScribe's AI capabilities are backed by Bedrock, a platform that incorporates pretrained models from both startups and Amazon. To address concerns regarding biases and accuracy, HealthScribe currently focuses on clinical notes for general medicine and orthopedics, allowing clinicians to review and finalize records in their EHR.

Amazon also emphasizes the security and privacy aspects of HealthScribe, highlighting that customer data is not retained after processing, data is encrypted, and inputs and outputs are not used to train Amazon's AI models. Although HealthScribe is not initially compliant with HIPAA, the U. S. law protecting personal health information, it is considered "HIPAA eligible" and can achieve compliance through agreements between customers and Amazon. Companies like 3M Health Information Systems, Babylon Health, and ScribeEMR are already using HealthScribe. Additionally, Amazon announced AWS HealthImaging, a service that simplifies the storage, transformation, and analysis of large-scale medical imaging data in the AWS cloud. HealthImaging offers dynamic pricing and rapid image access, potentially reducing the total cost of ownership for medical imaging storage. The service is now available in various AWS regions.