Apple Launches AI Health Study App With OpenEvidence

Apple launched the Apple Health AI Usage Study app with OpenEvidence, enrolling up to 200,000 US iPhone users to evaluate clinical AI queries.

Apple launched the Apple Health AI Usage Study on November 15, 2026, releasing a dedicated research app powered by clinical AI provider OpenEvidence to examine how consumers consult artificial intelligence for everyday health queries. The virtual clinical trial seeks up to 200,000 adult participants across the United States who own an iPhone 15 Pro or later running iOS 26 or higher.


Executive Summary

  • Study Name**: Apple Health AI Usage Study
  • Technology Partner**: OpenEvidence (clinical medical AI platform)
  • Enrollment Target**: Up to 200,000 adult iPhone users in the United States
  • Hardware Requirement**: iPhone 15 Pro or newer device running iOS 26+
  • Primary Source Announcement**: Apple Newsroom Official Health Study Release

Table of Contents


Technical Factsheet: Apple Health AI Research Architecture

The following table summarizes the operational parameters, data governance standards, and technical specifications of the study:

ParameterSpecificationVerification Standard
Enrollment Cap200,000 adult participantsGeofenced to United States residents aged 21 and older
Supported DevicesiPhone 15 Pro, iPhone 15 Pro Max, iPhone 16 seriesApple Intelligence hardware with A17 Pro or A18 silicon
Clinical AI EngineOpenEvidence clinical language modelTrained on peer-reviewed literature (NEJM, JAMA, Cochrane)
HealthKit IntegrationOptional read-only accessResting heart rate, step count, sleep stages, and lab records
Query Data Retention24-hour automatic purge on OpenEvidenceNo prompt data stored permanently or used for model training
Data De-identificationClient-side tokenizationPersonal identifiers stripped before secure cloud transmission
Study Duration12 to 24 monthsLongitudinal surveys combined with in-app interaction metrics

Clinical Citation Engine and HealthKit Integration

The core capability evaluated in the study is the translation of complex medical guidelines into accessible consumer language. Rather than relying on general-purpose conversational models that are prone to hallucinating medical treatments, the app routes inquiries through OpenEvidence. Every answer generated by the engine includes direct inline citations to peer-reviewed journals, including The New England Journal of Medicine, JAMA, Nature Medicine, and the Cochrane Database of Systematic Reviews.

Participants can optionally link their Apple Health records to the app. When authorized, the model evaluates personal metrics such as resting heart rate trends, sleep duration, and blood glucose logs to contextualize health advice. For example, a user asking about post-exercise recovery can receive recommendations that account for their actual heart rate variability over the preceding 48 hours.


Data Privacy Architecture and Retention Controls

Apple implemented strict architectural firewalls to protect participant medical data:

1. Zero Model Training: Neither Apple nor OpenEvidence will use participant health conversations, survey responses, or HealthKit metrics to train foundation models. 2. 24-Hour Ephemeral Storage: OpenEvidence automatically deletes participant queries and generated answers from its inference servers within 24 hours of generation. 3. Encrypted Cloud Pipelines: All telemetry sent to Apple research servers is stripped of names, email addresses, and IP identifiers, utilizing random participant tokens.

Developers exploring health data integration can review TeckScaler’s analysis of building secure browser medical utilities.


Frequently Asked Questions

Who is eligible to join the Apple Health AI Usage Study?

United States residents aged 21 or older who possess an iPhone 15 Pro or newer device running iOS 26 or higher can enroll through the Apple Research app.

How does OpenEvidence prevent medical hallucinations?

OpenEvidence anchors its responses in verified clinical publications, enforcing retrieval-augmented generation across indexed medical journals with mandatory footnote citations.

Can participants withdraw from the study at any time?

Yes. Users can revoke HealthKit access or leave the research trial instantly within the Apple Research app settings, which stops all future data transmission.


Explore further engineering and policy reporting from TeckScaler:


Published by the TeckScaler Editorial Engineering Team. We report original technical analyses, regulatory shifts, and system architectures across modern artificial intelligence.