(EHL Poster) AI-Powered Detection of Preventive Care Gaps Using Natural Language Guidelines

Preventive care often fails to deliver value at the point of care because patients cannot be screened in real-time against complex guidelines like eCQMs and USPSTF. Existing rule engines are static and require perfectly structured data, which is rarely available. To solve this, we propose an AI system that uses large language models to: (1) automatically parse guidelines into machine-readable logic, and (2) infer patient eligibility by analyzing both structured EMR data and free-text notes. This system continuously evaluates patient charts, flags eligible individuals in real-time during routine visits, and delivers actionable recommendations. The result is closed care gaps, improved value-based performance, and proactive risk reduction.