The new study and its objectives

Apple has announced in the United States the launch of the Health AI Study, a research program accompanied by an application distributed on the App Store to understand how people use artificial intelligence to navigate health matters. According to ANSA Tecnologia, the trial aims to measure the quality of the generated responses, the level of trust users place in them, and the actual usefulness of digital support in preparing for medical visits.

The project arises in a context in which, again according to the source, about one in three adults in the United States already turns to a chatbot to obtain health information. The novelty, therefore, is not the use of AI itself, but the attempt to evaluate it with a structured research setup: a dedicated app, a large audience, and an observation period that can last up to twelve months. The question of results remains open, as Apple has not yet set a publication date.

How the app works and who can participate

Access is reserved for 200,000 U.S. residents aged at least 21 who own an iPhone compatible with Apple Intelligence: from the 15 Pro model onward with iOS 26. Within the app, participants can ask questions about symptoms or recovery paths and, if they wish, share metrics collected by HealthKit, such as sleep, laboratory tests, and heart rate, to receive more targeted feedback.

To generate responses, the system relies on OpenEvidence, a platform designed for doctors that draws on peer-reviewed scientific literature, including the New England Journal of Medicine, Nature, and JAMA. The choice signals a difference from general-purpose text generators: the stated reference is not the undifferentiated web but peer-reviewed scientific publications, an element that, according to the source, should favor documented answers.

Privacy, data processing, and open issues

The technical structure of the processing is described in a precise sequence. Initial processing takes place directly on the device using Apple's foundation models, to remove identifying data; at a later stage, requests pass through a secure server managed by Ust Global and are deleted by OpenEvidence within 24 hours. Apple has stated that the information collected will not be used to train artificial intelligence models.

On the verification front, several aspects remain to be clarified. It is not known when the results will be published, with what methodology trust and usefulness will be assessed, or how any inaccurate answers on health topics will be handled, a sensitive area for anyone using the system independently. The available material also does not indicate whether the study includes a control group or a comparison with traditional sources. Until then, the initiative should be read as a declared experiment and not as a scientific validation of health chatbots: its relevance lies in the fact that Apple is collecting comparable data on a large scale, while conclusions about effectiveness remain to be verified.

Source and attribution: this original summary was prepared by AICronaca based on the report by ANSA Tecnologia ↗. Statements are attributed to the source; any informational limitations: The material is based solely on the ANSA Tecnologia news item and does not contain results, detailed methodology, a control group, or data on any clinical risks. Part of the original text consists of subscription messages and cookie policy, which were not used. The compatibility indicated is that reported by the source..