MIPS quality measures with OpenAI 4o Mini

Automating MIPS quality scoring with next-gen AI models​

​
The team developed a prototype using OpenAI 4o Mini and Retrieval-Augmented Generation (RAG) to automate quality measure calculations directly from patient notes. The system focused on ACEP22 (Pulmonary Embolism) and ACEP60 (Syncope), automatically detecting which MIPS measure applies based on unstructured clinical content. It then generates structured reports that can be seamlessly integrated into MIPS dashboards, streamlining compliance and reducing manual effort.

‍

Results

• 90%+ accuracy across tested notes
• Cost of processing: just $0.10 per note
• SaaS-friendly architecture with recurring revenue potential for enterprise deployment

Whats next?


The team aims to extend the solution to the full MIPS program—including Cost, Improvement Activities, and Promoting Interoperability—using a multi-agent AI approach.