From Patient to Practitioner: Why Patients Need a Voice in Governing the Artificial Intelligence Used in Bariatric and Diabetes Care

Artificial intelligence (AI) and other algorithmic tools are already part of bariatric and diabetes care: scores that predict diabetes remission after surgery, machine learning models that forecast glucose, body mass index (BMI) thresholds built into eligibility and coverage rules, remote monitoring platforms, and emerging surgical planning tools. The people most affected by these tools are rarely told they are being used and almost never take part in deciding how they are governed. In this Viewpoint, I write from both sides of that divide: as a patient who had gastric bypass surgery in 2018, later had excess skin removed, and now lives with type 2 diabetes that is well controlled, and as an AI governance consultant. I describe where AI touches this care pathway and identify 4 gaps that matter most from the patient's chair: transparency, consent, bias, and recourse. I then offer 5 recommendations that map onto frameworks health systems and developers are already adopting, namely the National Institute of Standards and Technology AI Risk Management Framework, ISO/IEC 42001, and the European Union Artificial Intelligence Act: patient representation in AI oversight, plain-language model disclosures, performance reporting for the subgroups patients belong to, notice at the point of care, and a clear route to question and appeal. Patients are not an audience for AI governance. We are part of the evidence it needs.