Apple Launches Health AI Usage Study for 200,000 US Adults
Apple launches the Health AI Usage Study with OpenEvidence, inviting as many as 200,000 US adults to test a conversational health tool and report how useful, understandable and trustworthy its answers are. Enrollment opened November 15 through a dedicated iPhone app.
The virtual study lets participants ask everyday health questions, view answers grounded in peer-reviewed clinical sources and optionally share selected HealthKit information. Apple says the research data will not train AI models, and the app is not intended to diagnose, treat or replace professional medical advice.
The Apple study is designed around four research questions:
- Which factors increase or weaken trust in health AI answers?
- How clearly do participants understand the information?
- Does optional health data make responses feel more relevant?
- Can the tool help people prepare for clinical visits?
Related Coverage
Apple Launches Health AI Usage Study for 200,000 Adults
The Apple Health AI Usage Study is open to eligible US residents aged 21 or older. Participants need an Apple Intelligence-capable iPhone, starting with the iPhone 15 Pro, and must run iOS 26 or later.
Apple created a separate study app instead of placing the experiment inside the Health app, Siri or its existing Research app. The Verge independently confirmed that the conversational tool itself does not use Apple Intelligence, even though enrollment requires compatible hardware.
Participants can ask about symptoms, general health, fitness, recovery and other everyday topics, then evaluate the answers. Apple is studying response quality, relevance, comprehension and subsequent health actions rather than testing whether an AI system can replace a clinician.
OpenEvidence Supplies Clinically Grounded Answers
OpenEvidence provides the underlying conversational AI technology. Its system generates responses from established medical sources that include The New England Journal of Medicine, the Journal of the American Medical Association and Nature, with linked clinical citations shown to participants.
The study joins Apple’s research design and device ecosystem with a platform already aimed at healthcare professionals. That pairing lets researchers examine whether a consumer can understand medical evidence presented in approachable language and recognize the limits of the answer.
The app also includes tools intended to help users prepare questions and summaries for visits with their own doctors. That handoff is central to the design: the study tests whether conversational AI can improve a clinical discussion while keeping diagnosis and treatment decisions with qualified professionals.
Optional HealthKit Sharing and Privacy Controls
Participants may choose to share individual HealthKit categories such as resting heart rate, activity, sleep or laboratory results. Apple says those choices are optional, allowing the study to compare general conversations with interactions informed by personal health measurements.
Apple states that study data will not be used to train AI models. It also says participants control whether to share HealthKit information and survey responses, while the app applies security measures intended to protect the data Apple maintains and the confidentiality of participant information.
Health-care publication Fierce Healthcare reported that questions are processed on the iPhone to remove identifying details before entering a secure study environment. They are separated from participant identifiers and screened again before OpenEvidence generates a response.
The app contains guardrails designed to detect certain concerning questions and direct the participant toward a provider or other resources. Those controls reduce some immediate risk, but Apple has not published accuracy rates for the detection system or the conversational answers.
Study Design Separates Research From Product Claims
The launch begins data collection rather than proving that conversational health AI improves outcomes. Apple has not announced a publication date, a final analysis plan or a timetable for turning any findings into features for the Health app or Siri.
A large enrollment target can reveal patterns across different users, questions and optional data combinations. It does not automatically establish clinical effectiveness: participation is limited to US adults with newer iPhones, which may produce a sample that differs from the broader population.
The useful evidence will come from retention, response ratings, comprehension measures, visit-preparation results and transparent reporting of failures. Until those results arrive, the Apple Health AI Usage Study is a controlled investigation of consumer behavior, not a validated medical product.
Illustrative feature photograph: “Personal Health Apps for Smartphones,” credited to Intel Free Press and sourced through Wikimedia Commons under CC BY 2.0. It does not show Apple’s study app or a current Apple interface. The photograph was cropped to 16:9 and resized.