Agentic AI-Generated vs. Clinician-Authored Hospital Course Narratives for Cardiology Hospitalizations: A Blinded Physician Evaluation
Main finding · Paired difference in overall quality (0-to-10 scale)Result — go to the resultDifference −0.15 95% CI −0.86 to 0.56Authors' conclusion
Hospital course narratives generated by the EHR-integrated agentic AI system did not differ significantly from clinician-authored narratives in physician-rated overall quality or preference. Lower completeness supports continued clinician review before use in the medical record.
Study context
- Population
- Inpatient cardiology encounters at Mayo Clinic, Rochester, Minnesota, admissions July 13 to 27, 2024inpatient cardiology encounters, 250 ratings
- Participants
- 50 encounters
- Exposure
- Artificial intelligence (AI)-generated and clinician-authored hospital course narratives
- Outcome
- Paired difference in overall quality (0-to-10 scale)
Methodological detailsrandomly selected encounters, five physician raters
- Randomly selected encounters
- Five physician raters
Main finding
As reported in the abstract
Difference
−0.15
95% CI −0.86 to 0.56
P = 0.627
By group
AI-generated
7.27
mean
Clinician-authored
7.42
mean
AI-generatedvs clinician-authored
difference −0.15 (95% CI −0.86 to 0.56); P = 0.627
Observational data: associations, not effects.
Further findings
Preference
| Outcome | Difference (95% CI) | P value | |
|---|---|---|---|
| ClarityNot significant after Bonferroni correction | 0.36 (0.07–0.65) | ||
| ReadabilityNot significant after Bonferroni correction | 0.72 (0.20–1.24) | ||
| CompletenessRemained significant after Bonferroni correction | −0.84 (−1.36 to −0.32) | adjusted p=0.012 | |
| AccuracyNot significant after Bonferroni correction | −0.35 (−0.66 to −0.05) | ||
| −2−1012Difference (linear scale) |
- Preference — odds ratio for clinician-authored preference
- 1.12 (95% CI, 0.71 to 1.78)
Original publication
Agentic AI-Generated vs. Clinician-Authored Hospital Course Narratives for Cardiology Hospitalizations: A Blinded Physician Evaluation. Circ Popul Health Outcomes. Published 6 November 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →