The India Decade

Case studyAI

How do tech giants study human interaction with health AI?

Technology companies are launching large observational studies to test how consumers trust, understand, and act on conversational medical advice.

By The India DecadeReviewed

The short answer

Major technology companies study human interaction with health artificial intelligence by launching large-scale observational research studies that track real-world user behaviour in real time. Rather than relying solely on laboratory benchmarks, platforms deploy conversational software to measure metrics like user trust, comprehension, response accuracy, and post-consultation health choices. Participants can optionally connect personal metric feeds, such as heart rate or sleep tracking, allowing researchers to evaluate how personal context alters communication effectiveness. These trials aim to establish clinical guardrails and design principles without using participant data to train underlying commercial models.

Up to 200,000 participants
Target enrollment
Apple and OpenEvidence, November 2026
21 years old in the United States
Minimum participant age
Apple, November 2026
iPhone 15 or later running iOS 26 or higher with Apple Intelligence enabled
Device requirement
Apple, November 2026
Zero study data used to train AI models
Data privacy commitment
Apple, November 2026
Peer-reviewed citations from NEJM, JAMA, and Nature
Literature source base
OpenEvidence, November 2026
More than 12 months
Study duration
Apple, November 2026
Query and response data deleted after 24 hours
Data retention
OpenEvidence, November 2026

What's happening now

Prompted by Apple launches new app to study how people use AI for health

In November 2026, Apple and OpenEvidence launched an experimental health chat application in the United States to evaluate how consumers interact with artificial intelligence for medical advice. Unveiled at the HLTH conference in Las Vegas by Apple vice president of health Sumbul Desai and OpenEvidence chief executive Daniel Nadler, the observational study aims to enrol up to 200,000 participants over a period of more than 12 months.

The app is available to US participants aged 21 and older using an iPhone 15 or later running iOS 26 with Apple Intelligence enabled. Rather than acting as a diagnostic tool, the system provides plain-language explanations of everyday health questions grounded in peer-reviewed research from The New England Journal of Medicine, Journal of the American Medical Association, and Nature. Users can choose to link HealthKit metrics—such as resting heart rate, sleep patterns, activity levels, and lab results—to obtain personalized context, rate the software's responses, and complete monthly surveys.

Privacy protocols for the trial require study data to be anonymised before transmission to a secure third-party cloud server, with Apple confirming that participant submissions will not be used to train AI models. OpenEvidence also stated that all query and response data is deleted after 24 hours. The project reflects growing industry focus on patient-facing AI tools, following statements from OpenAI earlier in 2026 that users were logging more than 300 million health queries every week.

How conversational health AI studies track user interaction

  1. Participant onboarding and eligibility

    Users aged 21 and older enroll through a free application on compatible devices, agreeing to study parameters and data-sharing preferences.

  2. Query submission and response processing

    Participants submit general health questions or symptom inquiries into a conversational interface backed by peer-reviewed literature.

  3. Optional biometric integration

    Users voluntarily link personal health metrics, such as resting heart rate or sleep logs, allowing the system to tailor contextual explanations.

  4. Citation generation and verification

    The system attaches direct links to peer-reviewed medical journals for every health claim made in the chat response.

  5. Real-time metric gathering

    Researchers capture user interaction patterns, survey answers, and follow-up choices to evaluate comprehension and trust levels.

Who it affects

While the initial Apple Health AI Usage Study focuses on enrolled users in the United States, conversational health tools are increasingly used across global markets, including India. In India, where doctor-to-patient ratios remain constrained in non-metropolitan areas, digital health applications and AI-driven chatbots are frequently used by consumers for preliminary symptom assessments before seeking formal medical care.

As platforms refine these models, understanding how users interpret conversational guidance affects millions of patients preparing for clinical visits worldwide. The research aims to establish baseline standards for clarity, ensuring that consumers do not mistake conversational summaries for formal diagnoses or actionable medical prescriptions.

What's being done, and is it working

  • Market

    Apple and OpenEvidence

    Launched a virtual research trial for up to 200,000 users to assess how consumers rely on conversational health tools.

    Announced in November 2026, long-term findings on user trust and comprehension remain under evaluation.

    Too early to say

  • Institution

    Department of Medicine at UCSF

    Evaluated the clinical utility of shared patient-doctor AI tools during primary care preparations.

    Dr Robert M. Wachter observed that shared trustworthy tools improve pre-appointment preparation and doctor-patient dialogue quality.

    Working

What you can do

  1. Verify any advice generated by health software with a qualified medical professional before changing treatment plans.

  2. Review app privacy permissions carefully to control which personal health metrics are shared with research studies.

  3. Look for explicit citations to peer-reviewed medical journals when reading AI-generated health explanations.

  4. Use health AI applications strictly to prepare questions for your doctor rather than as a substitute for a clinical diagnosis.

What we still don't know

  • How accurately can conversational AI models adapt to diverse regional languages and varied medical literacy levels worldwide?
  • Will regulatory bodies establish mandatory clinical evaluation standards for consumer-facing health AI applications?
  • To what extent do contextual biometric metrics improve patient outcomes compared to standard un-personalised health queries?

Questions readers ask

Is conversational health AI designed to replace doctors?

No. Research applications are designed to help users understand general health topics and prepare for doctor appointments, not to deliver clinical diagnoses or treatment decisions.

Is personal health data used to train artificial intelligence models in these studies?

Apple has stated that participant data collected in the Apple Health AI Usage Study will not be used to train underlying AI models.

Who is eligible to participate in Apple's health AI research study?

Enrollment is limited to individuals aged 21 and older in the United States who own an Apple Intelligence-enabled device running iOS 26 or later.

Sources

  1. HLTH26: Apple launches 200K-person study to examine how consumers use health AIFierce Healthcare · Target study size and launch announcements
  2. Apple launches an AI health chatbot as part of a research studyABC News · Details on consumer health study app and research objectives
  3. Apple Opens U.S. Study of Trust in Conversational Health AIUnite.AI · Detailed breakdown of study eligibility, HealthKit integration, OpenEvidence platform, and privacy controls
  4. Artificial Intelligence (AI) in Cardiovascular Medicine - OverviewMayo Clinic · Background on AI application in clinical medicine research

Updates

  1. Updated in November 2026 following the conference announcement by Apple and OpenEvidence, adding details on participant device criteria, study duration, and data deletion timelines.

Topics

The day’s reporting, once each morning. No advertising.

The news on this