Amazon has unveiled an automated healthcare claims processing pipeline using Amazon Bedrock and AWS HealthLake. The solution aims to cut down on manual processing time by leveraging AI to extract and validate data from paper-based forms. The system reduces the need for human oversight, ensuring accuracy through automated validation checks. The pipeline is designed to handle CMS-1500 claim forms uploaded to Amazon S3, triggering a series of automated workflows. The goal is to streamline the processing of healthcare claims, minimizing errors and improving efficiency in the industry.
The automated workflow begins when a healthcare provider uploads a CMS-1500 form to Amazon S3, which triggers AWS Lambda. The process starts with Amazon Bedrock Data Automation extracting structured data from the form using intelligent document processing. This data is then validated by an AI agent hosted on Amazon Bedrock AgentCore, which cross-checks the information against existing patient and provider records in AWS HealthLake. If all validations pass, the agent creates a standardized FHIR claim resource in HealthLake and generates summaries for both claims processors and patients. These summaries are delivered as Amazon SNS notifications, ensuring transparency and clarity in the claim status.
The solution uses Amazon Bedrock Data Automation to streamline generative AI development and automate workflows involving documents, images, audio, and videos. For document processing, Bedrock Data Automation combines traditional OCR, machine learning models, and generative AI to extract data accurately. Users can use Blueprints to specify what data to extract and how, with pre-built templates or custom configurations tailored to specific use cases. The output includes confidence scores and bounding box data for extracted fields and tables, ensuring a predictable JSON representation of the CMS-1500 form across its format variations.
Source: awsml
