CARE-AI framework sets responsible AI standard for health training
A JMIR Medical Education study details a consensus framework for safer AI use in medical education, research and patient care.
By Priya Raghavan · Science Reporter
3 min read
An international group has introduced the CARE-AI framework to guide responsible artificial intelligence use in health professions education, research and patient care. The framework matters because clinicians, teachers and health systems are already facing practical questions about privacy, bias, consent and human oversight as AI tools enter training and clinical work.
JMIR Publications said the Health CARE-AI framework, short for Contextual, Accountable, Responsible and Equitable Artificial Intelligence, was published in JMIR Medical Education. The study was led by Lyn K. Sonnenberg and colleagues and is described as a Delphi consensus study of principles for responsible AI across health professions education, research and care.
What is the CARE-AI framework?
The CARE-AI framework is a set of professional expectations for how health educators, researchers and clinicians should use AI. According to the study, it translates broad AI ethics principles into competencies, accountability practices and equity commitments that can be taught, supervised and built into institutions.
The researchers said the work was developed through a three-phase modified Delphi process involving 303 unique international participants. JMIR Publications reported that 96% of participants agreed or strongly agreed that the framework clearly defines professionalism expectations for AI across educational, technological and ethical needs.
The study places the framework in the gap between high-level policy and daily decision-making. It notes that groups including the World Health Organization and UNESCO have issued broad principles for AI, while clinicians and educators still need clearer direction on issues such as patient privacy, student assessment, algorithmic bias and data stewardship.
According to the study, AI creates challenges for clinical autonomy, trust between people in care settings and health equity that go beyond earlier digital and social media tools. The authors designed CARE-AI to give health programs and systems a way to connect values with everyday responsibilities.
What principles does it include?
The framework organizes 10 principles into four domains: values, competence, accountability and structural equity. The study says the values domain treats responsible AI use as both an individual and collective duty, with transparency, honesty and integrity expected in AI-assisted care and learning.
The competence domain calls for continuing, role-appropriate AI literacy and for people to maintain critical human judgment. In the study’s framing, AI should support clinical and educational decisions rather than replace the judgment of health professionals and educators.
The accountability domain treats AI as a present “third party” in health and learning interactions, according to the study. It also emphasizes legal, privacy and consent boundaries, along with ethical handling of data.
The structural equity domain focuses on identifying and reducing algorithmic bias, embedding equity into AI design and governance, and involving affected communities in co-design. The study also includes environmental and workforce sustainability within that domain.
How could health programs use it?
JMIR Publications said the CARE-AI framework is paired with an implementation guide and toolkit rather than standing alone as a policy document. The toolkit includes scenario-based uses across teaching, research and governance.
According to the study, medical schools, residency programs and health system leaders can use the materials to assess readiness, update curricula and audit AI deployments. The authors present the framework as a way to protect patient welfare, preserve trust and support equity as AI becomes more common in health education and care.
This story draws on original reporting from Medical Xpress.