Effect of Previsit AI-Generated Document Summaries on Consultation Time and Patient-Physician Communication - The AI-INFOCARE Randomized Controlled Trial.
Elida Benker-Hasani, Sven Richter, Clara Buszello et al.
✦ AI-curated · Sources linked
In 30 seconds
This single-center randomized controlled trial evaluated whether a quality-checked, AI-generated lay summary of medical findings before outpatient neurosurgical consultations affected communication and consultation time. Among 300 adults, the intervention group reported higher interaction quality than standard care (mean FAPI score, 4.42±0.51 vs. 4.25±0.54; difference, 0.17; 95% CI, 0.05 to 0.29) and had shorter median consultations (25 versus 33 minutes). Patient-rated autonomy support was also higher, while physicians rated their own autonomy support slightly lower. The authors call for further evaluation before routine adoption.
Key findings
- Mean patient-reported FAPI scores were 4.42±0.51 versus 4.25±0.54; difference, 0.17 (95% CI, 0.05 to 0.29; P=0.005).
- Median consultation time was 25 versus 33 minutes in the intervention and control groups, respectively (ratio of geometric means, 0.79; 95% CI, 0.70 to 0.88).
- Patient-rated autonomy support was higher (difference, 0.14; 95% CI, 0.06 to 0.23), while physician-rated autonomy support was lower (difference, -0.12; cluster-robust 95% CI, -0.21 to -0.04).
- Evidence for increased encounter difficulty was limited: difference, 0.09 (cluster-robust 95% CI, -0.03 to 0.21), compatible with no difference.
Why it matters
Patient-facing explanations of medical findings may help patients participate in consultations while reducing time spent in the encounter. The differing patient and physician ratings of autonomy support, and the uncertain signal for encounter difficulty, show that communication effects may not be uniform. The trial offers evidence about interaction outcomes, but does not establish effects on clinical outcomes or clinician workload.
What to keep in mind
The trial was conducted at a single center in an outpatient neurosurgical setting. Secondary-endpoint confidence intervals were unadjusted for multiplicity, and clinical outcomes and clinician workload were not established.