Apple launches Health AI study app for up to 200,000 U.S. iPhone users

Apple has opened enrollment in the Apple Health AI Usage Study, a virtual research study in the United States that gives participants a dedicated iPhone app for asking everyday health questions to a conversational AI. The company announced the study on its Newsroom on November 15, 2026, and said it will accept up to 200,000 eligible participants aged 21 and older who own an Apple Intelligence-enabled iPhone, which Apple defines as an iPhone 15 Pro or later running iOS 26 or later.
The study is built with OpenEvidence, a clinical AI platform that physicians already use to search the medical literature. Apple says the answers participants receive draw on established medical sources, including The New England Journal of Medicine, the Journal of the American Medical Association and Nature, and that responses come with linked, peer-reviewed clinical citations. Participants can also choose to share selected HealthKit data from their iPhone, such as resting heart rate, activity, sleep or lab results, so that the answers can take their own measurements into account.
Apple presents the project as research into how people actually use health AI, with a focus on what drives trust, the quality of the responses, how well people understand them and what they do with the information afterward. The company states that study data will not be used to train AI models, that participants decide what they share, and that the app is not intended to diagnose or treat any condition or to replace advice from a medical professional. Enrollment is open now through the study app on the U.S. App Store.
··········
WHO CAN JOIN THE APPLE HEALTH AI USAGE STUDY.
The eligibility rules are short, and every one of them is a hard requirement. The study is limited to the United States, so iPhone owners elsewhere cannot enroll even if their device and software qualify. The minimum age is 21, and Apple applies it to every participant regardless of whether the person already uses Apple's other health features.
The hardware requirement follows the Apple Intelligence line. Apple specifies an Apple Intelligence-enabled iPhone, meaning an iPhone 15 Pro or a later model. The standard iPhone 15 and iPhone 15 Plus do not support Apple Intelligence, so owners of those phones fall outside the study even though the models were released in the same generation as the iPhone 15 Pro. The software requirement is iOS 26 or later, which also covers phones already updated to newer releases.
........
Requirement | Apple Health AI Usage Study |
Country | United States only |
Minimum age | 21 |
iPhone | Apple Intelligence-enabled: iPhone 15 Pro or later |
Software | iOS 26 or later |
Maximum participants | Up to 200,000 |
App | Dedicated study app on the U.S. App Store |
HealthKit sharing | Optional, chosen by the participant |
Use of study data for AI training | No, according to Apple |
........
The cap of 200,000 participants sets the scale of the study, although Apple has not said how quickly it expects the study to fill or whether enrollment will close once the cap is reached. Apple also has not published a study end date, so participants cannot yet tell from the announcement how long they would be asked to stay enrolled.
··········
HOW THE APPLE HEALTH AI STUDY APP ANSWERS HEALTH QUESTIONS WITH OPENEVIDENCE.
The core of the app is a conversational interface for everyday health questions. Apple describes three main uses: asking routine questions about health, getting complex medical information explained in plainer language, and following linked citations back to the peer-reviewed research behind each answer. The citation model reflects how OpenEvidence works for clinicians, where answers are tied to published studies so the reader can check the underlying evidence.
OpenEvidence is a clinical decision support platform aimed at verified healthcare professionals in the United States, which is why the consumer study is a departure for it. In this study, the medical knowledge comes from OpenEvidence's literature base, while Apple runs the study, provides the iPhone app and handles the optional connection to HealthKit and the privacy controls it describes. The result gives ordinary iPhone users access to an answer engine that was built for doctors, wrapped in a format meant for people without medical training.
The optional HealthKit connection adds personal context to those answers. A participant who shares resting heart rate, activity, sleep or lab results can ask questions about their own numbers instead of generic ones, for example how a recent lab value compares with typical reference ranges or how sleep data might relate to a question about fatigue. Apple says one of the study's aims is to learn whether this kind of personal data makes the responses more relevant and useful.
The study also looks at a specific use case that many people already attempt on their own with chatbots: preparing for an appointment. Apple says it will explore how health AI can help participants get ready for visits with their own doctors, which in practice means organizing questions, understanding terminology and arriving with a clearer picture of their own data.
··········
APPLE HEALTH AI STUDY PRIVACY: HEALTHKIT SHARING, TRAINING AND SAFETY GUARDRAILS.
Privacy is the area where Apple has given the most specific commitments, and it is likely to decide whether many eligible users enroll. The company makes two central statements in its announcement. Study data will not be used to train AI models, and participants control whether they share HealthKit metrics and whether they share survey responses.
........
Area | What Apple says about the study |
AI model training | Study data will not be used to train AI models |
HealthKit data | Shared only if the participant chooses; examples include resting heart rate, activity, sleep and lab results |
Survey responses | Shared only if the participant chooses |
Safety guardrails | Designed to detect certain concerning questions and suggest following up with a provider or other resources |
Medical role | Not intended to diagnose or treat conditions or to replace professional medical advice |
........
The safety guardrails matter for a consumer health product because people often ask chatbots about symptoms that need urgent care. Apple says the app includes guardrails that detect certain concerning questions and point the participant toward a healthcare provider or other resources. The company has not published the list of situations that trigger these responses, and it has not described how the system behaves when a question falls between routine and urgent.
Apple's announcement does not describe the full data flow between the iPhone, Apple's systems and OpenEvidence, and it does not state retention periods for questions and answers. Participants who want those details will need to read the informed consent documents in the study app before enrolling, which is the stage where research studies normally set out data handling, retention and withdrawal rights.
··········
[[ADPLUS_300x250_1]]
WHAT APPLE IS MEASURING IN THE HEALTH AI USAGE STUDY.
The research questions are framed around behavior rather than diagnostic accuracy. Apple says the study will examine what drives user trust in conversational health AI, the quality of the AI-generated responses, how well participants understand the information, and what everyday health actions they take afterward. Participants contribute by using the app and, if they choose, by sharing survey responses.
Sumbul Desai, M.D., Apple's vice president of Health and Fitness, said in the announcement that "responsible innovation starts with scientific rigor and clinical grounding," and described the study as a way to inform the responsible design of health AI. Robert M. Wachter, M.D., of the University of California, San Francisco, who is quoted in Apple's release, warned that "the tools they use are being built and tested in separate worlds," a point about the gap between how health AI is developed and how people actually use it.
That framing explains the choice of partner. A clinician-grade evidence engine tested on a large consumer population lets Apple observe where answers that are correct on paper still fail in practice, for example when a participant misreads a well-sourced explanation or trusts an answer more or less than its evidence supports. These are questions that accuracy benchmarks on curated medical exams do not answer, and they require real usage at scale.
··········
HOW TO JOIN THE APPLE HEALTH AI STUDY ON AN IPHONE.
Enrollment runs entirely through the iPhone. Based on Apple's description of the study, the process works as follows.
1. Confirm that the iPhone is an iPhone 15 Pro or a later model and that it runs iOS 26 or later, with Apple Intelligence supported on the device.
2. Download the Apple Health AI Study app from the U.S. App Store.
3. Review the study information and complete the informed consent process inside the app.
4. Choose whether to share HealthKit data types such as resting heart rate, activity, sleep or lab results, and whether to share survey responses.
5. Start asking health questions and complete the study surveys if you have chosen to share your responses.
Several practical details are not in Apple's announcement. The company has not stated how long the study runs, whether participants receive any compensation, or how withdrawal works, although research apps normally explain withdrawal in the consent documents. Readers who already use Apple's own health tools can compare the study with the features Apple is adding to the main Health app, covered in our report on the redesigned Apple Health app in the iOS 27.3 beta.
··········
WHERE THE HEALTH AI STUDY FITS IN APPLE'S HEALTH AND SIRI AI PLANS.
The study arrives while Apple is expanding its AI features across the iPhone, including Siri AI, which is still unavailable in the European Union, as described in our coverage of Siri AI's new languages in iOS 27.2. Health has been one of the categories where Apple has moved most cautiously with generative AI, and the decision to launch a research study with an outside clinical partner, instead of shipping a health chatbot inside Siri or the Health app, is consistent with that caution.
Apple has not said that the study app will become a consumer product, and it has not announced a health assistant for the Health app. The design of the study, with a defined population, informed consent, optional data sharing and an explicit statement that study data will not train models, resembles Apple's earlier research programs more than a product beta. The findings could still shape future features, since Apple says the goal is to inform the responsible design of health AI.
For OpenEvidence, the study opens a consumer channel for a platform whose own app is offered to verified U.S. clinicians. The company's involvement suggests that citation-based answers are the format Apple considers acceptable for health questions, in contrast with general-purpose chatbots that answer without pointing to specific clinical sources.
··········
APPLE HEALTH AI USAGE STUDY: CONFIRMED FACTS AND OPEN QUESTIONS.
The confirmed elements are clear. Apple announced the Apple Health AI Usage Study on November 15, 2026, for up to 200,000 U.S. participants aged 21 and older with an iPhone 15 Pro or later on iOS 26 or later. Answers come from OpenEvidence with linked peer-reviewed citations, HealthKit sharing is optional, study data will not be used to train AI models, safety guardrails redirect certain concerning questions to providers, and the app is explicitly not a diagnostic or treatment tool.
The open questions concern the terms of participation and Apple's plans after the study. Apple has not published the study's duration, any compensation, the detailed data flow to OpenEvidence or retention periods for questions and answers. It has not said when results will be published or whether a consumer health assistant will follow. Until those points are documented, the study is a large, opt-in experiment in how Americans use clinician-grade AI for their own health questions, and eligible iPhone owners can judge the trade-off by reading the consent documents before sharing any HealthKit data.
··········
FOLLOW US FOR MORE.
·····
DATA STUDIOS
·····
datastudios.org




