Physicians in value-based contracts are doing two jobs at once: delivering care and proving it. Risk adjustment, quality reporting and payer-specific documentation all land on the same clinicians already spending 5.3 hours in the EHR for every eight hours of clinic.
Phyx Primary Care, a non-profit innovation lab that takes no funding from the companies it evaluates, set out to test whether AI could absorb that second job. The evaluation was conducted by Edmund Billings, MD, and Steven E. Waldren, MD, MS, chief medical informatics officer of the American Academy of Family Physicians.
Across 120 physicians at 19 practice sites in 13 states, the study measured what changed after at least 30 days of using an AI assistant built into the point of care. Reported burden dropped. Trust in the tool ran high. And the participating organizations reported measurable movement in the contract metrics that decide whether a practice stays solvent.
Inside the report:
- Why 92% of physicians trusted the assistant’s hierarchical condition category suggestions
- How self-reported clinical review burden fell 24% and burnout fell 32%
- What the 8% who would not recommend the assistant said about where the work went instead