Apple launched the Apple Health AI Usage Study with OpenEvidence on November 15, seeking up to 200,000 U.S. participants to examine trust, comprehension and everyday engagement. The interesting question isn’t whether AI can answer health questions, but whether people understand those answers.

“We’re studying what builds user trust, how people comprehend responses, the quality of responses,” Dr. Sumbul Desai, Apple’s vice president of Health and Fitness, said onstage during the partnership announcement at the HLTH conference in Las Vegas, STAT reported.

Apple’s announcement describes an Apple-designed experience using OpenEvidence’s clinical AI, naming The New England Journal of Medicine, JAMA and Nature among its sources. Eligibility starts at age 21 and requires an Apple Intelligence-enabled iPhone, an iPhone 15 Pro or later, running iOS 26 or later.

ABC News reported on November 15, 2026, that participants download a dedicated app and complete informed consent; full OpenEvidence access currently remains restricted to healthcare professionals. TechCrunch’s November 15 report says participants will receive monthly surveys. STAT also covered the health-chat-app experiment that day.

The doctor is still in the picture

Apple will explore whether the tool helps participants prepare for visits with their own clinicians.

“Patients and clinicians are both turning to AI, but the tools they use are being built and tested in separate worlds.” That’s how Dr. Robert M. Wachter, UCSF’s professor and chair of the Department of Medicine, describes the divide in Apple’s announcement.

Citations aren’t a substitute for care. Apple acknowledges that consumer health tools powered by large language models can produce “incomplete or unclear” responses. It says the app isn’t intended to diagnose, treat or replace professional advice. ABC News reports that it doesn’t provide live medical monitoring.

Apple says study data won’t train AI models, and participants control whether they share HealthKit metrics or survey responses. According to ABC News, Apple says identifying information is removed before questions reach OpenEvidence, which deletes question-and-response data within 24 hours. Apple also plans to make the research broadly available.

Those are research aims and safeguards, not findings demonstrating better understanding, better appointments or better health decisions.

For independent practices, a hypothetical patient-acquisition and retention approach follows: explain how patients can bring AI-informed questions into a human conversation, then make the follow-up step clear. That’s an inference about trust-building communication, not an outcome Apple has established.

The chatbot can supply an answer. Apple still has to study what happens after someone reads it.

Sources