AI-CLINIC: A Prospective Multisite Feasibility and Safety Study of AI-Assisted Pre-Consultation History-Taking in Outpatient Cardiology

An accurate medical diagnosis depends on the doctor getting a complete picture of the patient's health story, but clinic visits are short. Before some cardiology appointments, patients may be able to speak with an artificial intelligence (AI) "interviewer" - through a secure web page, usually at home - about their symptoms, medications, allergies, medical history, and what they want their doctor to understand. The AI interviewer creates a written summary for the cardiologist. It does not make diagnoses, give treatment advice, or replace the doctor: the cardiologist still takes and verifies the medical history in the usual way, and nothing written by the AI goes into the medical record without the doctor's review. In this study, 60 adults attending a first or new cardiology consultation at London Health Sciences Centre University Hospital or the London Cardiac Institute will be invited to use the AI interviewer before their appointment. Patients who prefer not to take part receive usual care. We will measure how many participants complete the interview and have a summary ready before their visit, how long the interview and the visit take, whether patients find the interviewer easy to use and feel able to explain their concerns, whether cardiologists find the summaries useful and accurate, and whether any important errors or omissions occur. Every summary is checked by the treating cardiologist, and independent physicians review all concerning cases plus a random sample. Results will show whether this workflow is feasible and safe enough to justify a larger study.

Inclusion Criteria: 1. Age 18 years or older. 2. Scheduled for a first or new in-person outpatient cardiology consultation at a participating site. 3. Able to provide free and informed consent. 4. Able to understand the approved English Letter of Information and consent discussion; the interview and study questionnaires are conducted in English. Additional languages may be added through a future amendment with translated, Board-approved materials. 5. Able to interact with the interviewer using a personal device or a clinic-provided iPad, with reasonable technical assistance if required. Exclusion Criteria: 1. Unable to provide informed consent. 2. Urgent or medically unstable presentation for which pre-consultation research procedures would be inappropriate or could delay care. 3. Severe cognitive, hearing, speech, or other communication limitation preventing meaningful voice interaction despite reasonable assistance. 4. Appointment cancelled, converted to a virtual/telephone visit, or otherwise no longer eligible before the intervention opportunity. 5. Previous participation in AI-CLINIC.