Software

Flo Health Uses Amazon Bedrock to Cut Medical Review Time by 60%

Flo Health reduced medical content review time by 60% using Amazon Bedrock, tripling content throughput without expanding its medical team.

Image: AWS Machine Learning

Flo Health, a digital health company, has implemented an AI-powered medical content review system built on Amazon Bedrock, significantly improving efficiency. The system, which integrates AI Judges and Retrieval Augmented Generation (RAG), allows the company to scale content production while maintaining rigorous medical accuracy standards. According to Flo Health, the solution reduced review time by 60% and tripled content throughput without increasing the size of its medical team. This marks a major shift in how the company handles content validation, leveraging AI to support human experts rather than replacing them.

The AI Review system operates through a three-layer validation process, first checking content against internal guidelines, then against trusted external medical sources, and finally being reviewed by human experts. Each stage is designed to flag potential issues, suggest corrections, and streamline the fact-checking process. Flo Health created a set of AI Judges, each tailored to specific review dimensions such as medical accuracy, legal compliance, and brand style. These Judges are trained with examples and tested with different models to ensure they meet quality thresholds while balancing cost and performance. The approach allows for independent improvements to each Judge without affecting the others, preventing regression in the system.

The implementation of the AI Review system is part of a broader effort to integrate Amazon Bedrock into Flo Health’s content pipeline. The company adapted MACROS patterns introduced by Amazon to its Contentful CMS, enabling efficient content chunking and review. By splitting content into smaller pieces and running specialized prompts for each review dimension, Flo Health ensures detailed feedback and targeted revisions. The system also allows content editors and medical experts to review, select, or edit AI-generated suggestions directly within their workflow, maintaining human oversight while leveraging AI for efficiency. This approach reflects a strategic balance between automation and expert input, ensuring accuracy and scalability.

Source: awsml

Key points

Flo Health reduced medical content review time by 60% using Amazon Bedrock.
The AI Review system tripled content throughput without expanding its medical team.
The system follows a three-layer validation process, checking content against internal guidelines, trusted external sources, and human experts.
Flo Health created AI Judges for specific review dimensions like medical accuracy, legal compliance, and brand style.
Each AI Judge is tested with different models to meet quality thresholds while balancing cost and performance.
The AI Review system integrates MACROS patterns from Amazon into Flo Health’s Contentful CMS.
Content is split into smaller pieces, with specialized prompts for each review dimension to ensure detailed feedback.

Source: AWS Machine Learning Read the original →

WRITTEN BY

Theo Almeida

AI Software & Developer Tools

Theo covers AI software, developer tools, frameworks, and the platforms builders use every day.