FastOMOP: Multi-Agent Architecture for Reliable RWE Generation on OMOP CDM
FastOMOP is a new open-source framework designed to make it easier to create real-world evidence (RWE) from data using the OMOP Common Data Model (CDM). This model, managed by OHDSI, combines electronic health records from almost one billion patients in 83 countries. Right now, producing RWE requires manual work and specialized knowledge in several areas like clinical science and technology. While large language models and multi-agent systems show promise, they can introduce unexpected behaviors and safety risks. FastOMOP includes three key layers—governance, observability, and orchestration—to ensure that RWE generation is flexible, secure, and can be audited, aiming to establish a reliable foundation for agentic RWE.
Key facts
- FastOMOP is an open-source multi-agent architecture for RWE generation on OMOP CDM data.
- OMOP CDM harmonizes electronic health records of nearly one billion patients in 83 countries.
- Current RWE generation is a manual process requiring clinical, epidemiological, and technical expertise.
- LLMs and multi-agent systems have shown promise but introduce emergent behaviors, coordination failures, and safety risks.
- No existing infrastructure ensures agentic RWE generation is flexible, safe, and auditable across the lifecycle.
- FastOMOP separates three infrastructure layers: governance, observability, and orchestration.
- The architecture addresses the gap in governing agentic systems for clinical tasks.
- FastOMOP is introduced in a paper on arXiv (2604.24572).
Entities
Institutions
- Observational Health Data Sciences and Informatics (OHDSI)
- Observational Medical Outcomes Partnership (OMOP)