This cross-sectional online study in France compared the performance of AI versus 204 general practitioners (GPs) in collecting patient medical histories across 8 categories (e.g., chronic diseases, allergies, family history of cancer) during simulated initial consultations.
Across 942 patient histories, AI outperformed GPs in history collection — particularly for older patients and in categories such as drug allergies and family history of cancer. Obstetric history and drug allergies were among the most relevant factors for accuracy.
The study was conducted entirely online, raising concerns about ecological validity compared to real clinical encounters. The specific AI tool(s) used are not described in the abstract, limiting reproducibility. No long-term patient outcomes or diagnostic accuracy data were reported.
AI may reliably supplement or exceed GP performance in structured history-taking, especially for older patients and allergy/family history categories. Clinicians should consider AI-assisted history tools while remaining attentive to ethical and training implications for the doctor-patient relationship.
Explore related topics
