The News, Explained

On November 15, Apple announced a U.S. study with OpenEvidence on how people use health AI. Participants can ask everyday health questions in a study app. Apple will look at how well people understand and trust the answers, and how that information relates to what they do about their health. It also plans to study whether the app helps people get questions ready for a medical visit. Source

The study is open to up to 200,000 eligible people in the United States who are at least 21. They need an iPhone that supports Apple Intelligence and runs iOS 26 or later. The app is free in the U.S. App Store and lists English as its language. The store’s 18+ label is a content rating; the study’s enrollment minimum is 21. Source Source

Answers link to peer-reviewed clinical references. HealthKit is Apple’s system for letting apps read health records that users give them access to. Participants can choose whether to share items such as resting heart rate, activity, sleep, or lab results. Apple wants to see whether using those records makes answers more relevant and easier to understand. Source Source

Sharing health metrics and survey responses is optional. Apple says study data will stay out of AI model training. The app is a research tool for exploring health information. Apple excludes diagnosis and treatment from its purpose and says it designed the app to direct people to a care provider or other resources for certain concerning questions. Source Source

OYOPICK’s Take

We want to see the study evaluate two things separately: whether an answer feels more personal when it uses someone’s records, and whether it is actually more accurate and easier to understand. An answer that refers to a sleep record can sound tailored to the reader. If the app shows which parts draw on that record and which offer general information, people may have a clearer idea of how much weight to give it.

Linking to references is a good start. If Apple shares results, we hope it shows whether people understood what those links support and where an answer’s limits are, and whether they were able to prepare more specific questions for a visit. The announcement sets out to investigate those effects.