Performance of artificial intelligence in the collection of patient history in general practice.

OBJECTIVE: Automating administrative tasks, such as compiling a patient's medical history, could help general practitioners in their daily work. AI performance has improved in recent decades, but skepticism among professionals limits its use in medical practice, due to fears of gaps and biases. This study attempts to evaluate the effectiveness of AI in recording patient histories compared to general practitioners. DESIGN: Cross-sectional study. SITE: Online study in France. PARTICIPANTS: French general practitioners recruited online. INTERVENTION: We compared the performance of a general practitioner with that of AI in collecting a patient's history during an initial consultation. MAIN MEASUREMENTS: Medical histories were classified into categories: long-term chronic diseases, medical, surgical, obstetric, occupational, drug allergies, other allergies, and family history of cancer. RESULTS: With 204 general practitioners, we obtained 942 patient histories. AI was more effective than traditional physicians in collecting patient histories, especially in older age groups and in categories such as allergies and family history. The relevance and reliability of those histories depend on the age and type of history reported. Older patients, obstetric history, and drug allergies were the most relevant factors. CONCLUSION: AI is expected to become a major player in the healthcare sector due to its efficiency and accuracy. However, there is still debate about its potential to replace doctors in tasks such as diagnosis, prevention, therapy, or patient information. The ethics of data collection and AI remain a subject of debate, with the importance of training healthcare professionals to ensure they respect medical and patient ethics. The study highlights the importance of integrating new technologies into general medicine and the doctor-patient relationship in the profession.