OpenAI has announced the launch of ChatGPT Health, the result of work with more than 260 physicians across 60 countries and multiple specialities, designed to help support 230 million people who seek health and wellness guidance every week.
ChatGPT Health aims to help people understand test results, prepare for appointments, and access advice on areas such as diet and workout routines, OpenAI suggests. Medical records and wellness apps can be securely connected to provide it with access to personal health information to improve responses, with emphasis on “supporting, not replacing, care from clinicians”.
Work with clinicians has focused on understanding what makes answers to health questions helpful or potentially harmful, OpenAI shares, with feedback provided on model outputs more than 600,000 times in 30 different areas of focus. “This collaboration has shaped not just what Health can do, but how it responds: how urgently to encourage follow-ups with a clinician, how to communicate clearly without oversimplifying, and how to prioritise safety in moments that matter,” it states.
A HealthBench assessment framework has been created to evaluate responses using physician-written rubrics that reflect how clinicians judge quality in practice, designed around principles such as safety, clarity, and appropriate escalations of care.
Additional layered protections designed specifically for health, including purpose-built encryption and isolation, are reportedly a feature of the new health platform, with users starting health-related conversations in ChatGPT being advised to move into Health for these additional security features.
Whilst Health has been assigned its own space within ChatGPT to allow chats and files to be stored separately, information from non-health related chats can be used to inform health discussions where relevant. Health information will not be flowed back into non-health related chats, however, and cannot be accessed outside of the Health space. OpenAI also assures that health conversations will not be used in training its foundation models. and offers users the opportunity to enable multi-factor authentication to further protect their information.
At the moment, medical record integrations and some apps are only available to US users. A small group of early users outside of the European Economic Area, Switzerland, and the UK, are being invited to access the experience to offer feedback, with access to be expanded to all web and iOs users in the coming weeks. Those interested in getting access as soon as it becomes available are invited to sign up to a waiting list.
Wider trend: AI in health and care
We were joined for a practical HTN Now webinar taking a deep dive into AI in health and care by expert panellists Peter Thomas, chief clinical information officer and director of digital development at Moorfields Eye Hospital; Sally Mole, senior digital programme manager – digital portfolio delivery team at The Dudley Group; and Ananya Datta, associate director of primary care digital delivery, South East London ICS. The session shared approaches, best practices, challenges, successes and learnings for the practical implementation of AI technologies across health and care, with our panel offering insight into current work, future plans, and ongoing collaborations in areas such as Ambient AI.
The Medicines and Healthcare products Regulatory Agency has issued a call for evidence on the regulation of AI in healthcare, asking members of the public, clinicians, industry, and healthcare providers to share their views on the modernisation of AI rules, keeping patients safe as AI evolves, and the distribution of responsibilities between regulators, companies, healthcare organisations, and individuals. In particular, views are sought on the UK’s current framework for regulating AI in healthcare and how it may need to be improved to “ensure fast access to safe and effective AI medical devices”, as well as on approaches to check safety once AI medical devices are in use, and how responsibility and liability are split between different parties involved in their deployment.
The US Department of Health and Human Services has published an AI strategy and accompanying compliance plan, to utilise leading technologies to enhance efficiency, foster American innovation, facilitate data and best practice sharing, standardise risk practices, and embrace a “try-first” culture. In its compliance plan, HHS identifies opportunities across several domains to minimise barriers to AI implementation, including in funding and procurement, IT infrastructure, product development, operations and maintenance, data management and sharing, cybersecurity, and workforce development.
