Patient and clinician engagement with generative artificial intelligence (GenAI): A scoping review of implications for patient-centered communication.
Jessica Hahne, Brian D Carpenter
PMID 41570489WHAT IT FOUND
GenAI may save clinicians documentation time and help patients understand test results, but its answers vary in accuracy, relevance and reading level.
Patients may expect clinicians to check AI output, so use it cautiously.
Key findings
01GenAI draft notes and messages did not consistently save clinician time: two studies found significant time saved, one found no difference, and one found increased time.
02ChatGPT-family models answering patient questions showed accuracy from 54.5% to 93.9% across studies.
03Without explicit prompts to lower reading level, LLM output was often written too high for the average American to understand.
STILL TO COME
How it was doneWhat they foundWhat it means for RNs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The review did not assess quality or risk of bias for each included study. Most included studies were cross-sectional, had small samples, and used heterogeneous GenAI models and healthcare settings, so effects are hard to compare or generalise. Many studies did not report model version or how prompts were entered, making results hard to reproduce. The review mapped existing studies rather than testing GenAI directly, so it cannot show that GenAI improves patient-centered communication or outcomes. Only four included studies tested interventions, and one rehabilitation finding came from a single study.
Declared interests
One author, Jessica Hahne, reported financial support from the National Institute on Aging. Other authors declared no known competing financial interests or personal relationships.
The easy way to misread this
Do not read this review as proof that GenAI improves patient-centered communication or clinical outcomes. Most included studies were cross-sectional, samples were small, and the review did not assess study quality or risk of bias.