Related Experiment Videos
Effect of Previsit AI-Generated Document Summaries on Consultation Time and Patient-Physician Communication - The
Elida Benker-Hasani1, Sven Richter1, Clara Buszello1
1Department of Neurosurgery, Faculty of Medicine and University Hospital Carl Gustav Carus, TUD Dresden University of Technology, Dresden, Germany.
Background:
Limited health literacy, clinical jargon, and time-constrained consultations often impair patients' understanding of medical findings. We evaluated whether AI-generated, quality-checked lay language summaries provided before consultations improve communication-related outcomes.
Methods:
In this single-center, randomized trial in an outpatient neurosurgical setting, 300 adults were randomly assigned 1:1 to standard care with or without a quality-checked, AI-generated layperson summary of their medical findings (Simply Onno) before consultation. The primary end point was patient-reported quality of physician-patient interaction (Fragebogen zur Arzt-Patient-Interaktion [FAPI]). Secondary end points included autonomy support (Health Care Climate Questionnaire [HCCQ]), encounter difficulty (Difficult Doctor-Patient Relationship Questionnaire [DDPRQ-10]), and consultation duration (minutes).
Results:
The mean FAPI score was higher in the intervention group (4.42±0.51 vs. 4.25±0.54; difference, 0.17; 95% confidence interval [CI], 0.05 to 0.29; P=0.005; Hedges' g=0.33), driven by shared decision-making (difference, 0.29; 95% CI, 0.15 to 0.42), with no evidence of a difference in perceived empathy. Patient-rated autonomy support was higher (difference, 0.14; 95% CI, 0.06 to 0.23), whereas physicians rated their own autonomy support slightly lower (difference, -0.12; cluster-robust 95% CI, -0.21 to -0.04). Median consultation time was 25 versus 33 minutes in the intervention group versus the control group, respectively, an approximate 21% relative reduction (ratio of geometric means, 0.79; 95% CI, 0.70 to 0.88). For encounter difficulty (DDPRQ-10), we found potential evidence of a small increase (difference, 0.09; cluster-robust 95% CI, -0.03 to 0.21). Secondary end point confidence intervals were unadjusted for multiplicity.
Conclusions:
AI-generated summaries provided before outpatient neurosurgical consultations were associated with better patient-reported interaction quality and autonomy support and shorter consultations. Physicians rated their own autonomy support as slightly lower in the intervention group; evidence for increased encounter difficulty was limited, with confidence intervals compatible with no difference. These findings support further evaluation of quality-checked, patient-facing document translation; effects on clinical outcomes and clinician workload require confirmation before routine adoption.