Medical-AI trial: easing the documentation burden with AI (Evidently) — a randomized test against clinician burnout (NCT07498582)

In health care, documentation can take as much time as care itself. Chasing electronic health record (EHR) review and entry, with work spilling past the visit and after hours, is a major driver of clinician burnout. A growing focus is clinician-facing AI that helps the doctor, not the patient, easing documentation by summarizing record information. This study evaluates one such tool, Evidently.

Per the registry summary, this randomized evaluation tests whether an AI documentation-support tool that creates brief summaries of existing record information to support routine work can help outpatient specialists spend less time reviewing and documenting.

It examines how using the tool affects clinician workload, time in the EHR, and overall documentation experience, aiming to understand whether such AI can improve efficiency and reduce burden in outpatient specialty practice. Participants are UNC (University of North Carolina) outpatient specialists, and the primary outcome is the change in clinician cognitive load. 128 clinicians; completed.

What is distinctive is that this is not about AI diagnosing or treating disease but about AI supporting how clinicians work. Rather than competing on diagnostic accuracy, setting clinician cognitive load and EHR time, measures of workability, as the outcome reflects a view that AI value is measured not only by clinical outcomes but by clinician well-being and sustainability.

Easing clinician burden can ripple to overall care quality by curbing turnover and burnout. That said, this study evaluates load, time, and experience; it does not broadly establish improvement in patient outcomes or the tool effectiveness.