Dive Brief:
- Apple is launching a virtual research study — and an associated app — to understand how people interact with conversational health artificial intelligence, the company announced Sunday at the HLTH conference in Las Vegas.
- The study, which is being conducted in partnership with clinical AI platform OpenEvidence, will examine what drives consumer trust, the quality of health AI responses and whether conversational AI can drive behavior change, according to Dr. Sumbul Desai, Apple’s vice president of Health and Fitness.
- Apple could use the findings to inform its development of AI products for healthcare. However, study data will not be used to train AI models, and users will have to opt into sharing their personal data, Apple said.
Dive Insight:
Apple’s new study is open to up to 200,000 eligible participants age 21 and older across the U.S., per a press release. It will be conducted through an app, where users can ask questions about their health and explore medical information. Participants will also have the option to share their personal data, such as resting heart rate, activity, sleep or lab results, to personalize their interactions.
The study was designed by Apple, but uses AI developed by OpenEvidence, Desai said during Apple’s session on HLTH’s main stage. OpenEvidence offers an AI-powered medical search tool that’s used by almost two-thirds of U.S. physicians, the company told NBC News earlier this year.
“We’re studying what builds user trust, how people comprehend responses, the quality of responses, and we wanted to work with a clinically grounded partner, and that is OpenEvidence,” Desai said.
Healthcare products powered by large language models often give incomplete or confusing answers. Apple is hoping that the research can inform how to present complex information in a clearer way, the company said. The app could also help consumers prepare for visits with their doctors.
OpenEvidence CEO Daniel Nadler, who joined Desai onstage, said that patients’ interactions with physicians are so short — frequently just 10 or 15 minutes — that conversations might fall through the cracks. The new app could help communicate users’ health data and questions to bridge that divide, Nadler said.
Americans remain leery of AI in healthcare, but they’re increasingly accepting of the technology when it comes to seeking out and absorbing health information. Roughly one-third of U.S. adults have used an AI chatbot to talk about their health, according to the KFF.
Healthcare companies have been racing to weave generative AI into their products, and wearables makers are no different. Companies like Google and Oura are integrating AI-based coaches into their products, with the goal of translating raw metrics into actionable guidance for users. As specialized LLMs become more advanced, they’ve had more success in answering clinical questions.
Still, concerns remain about the potential for hallucinations, incomplete answers or a lack of safety guardrails.
Apple said its new app will be able to detect worrying questions and direct a consumer to an outside provider or other resources if needed.
It’s the latest health research initiative for Apple, which has been testing how well its Apple Watch can track mobility, heart health, hearing changes and fertility. Last year, the company unveiled the Apple Health Study to explore if its devices can influence a wide range of health and disease outcomes, in collaboration with Brigham and Women’s Hospital.