The standard and accompanying framework aim to provide clear guidance and expectations for health AI tools that interact with patients.
A group of health system, patient safety and technology leaders has launched an operational standard focused on patient-facing health AI tools.
The AI Care Standard provides guidelines for safe and responsible patient-facing AI, which the developers define as “AI that communicates directly with patients, or AI that mediates, generates, or meaningfully shapes provider communication with patients.”
The standard’s launch comes as AI-generated patient communication through patient portals, chats and care navigation and outreach tools become increasingly common. While these tools can support patient education and potentially fill patient experience gaps, there is a lack of standardized safety guidance for these types of communication. In fact, misuse of patient-facing AI chatbots topped ECRI’s annual list of the biggest health technology hazards this year.
The standard aims to address the lack of guidance through 10 core pillars spanning safety, equity, governance and real-world usability. Some of the pillars include AI systems that respond appropriately to psychological, emotional, and situational aspects of interactions, adapt to individual needs to build trust and empower patients to understand their health while respecting clinical boundaries.
Developed by the PatientAI Collaborative cohort, the goal of the standard is to provide clear and enforceable expectations for patient-facing AI. The standard was developed through interviews with over 20 health system, policy, safety, patient experience and AI leaders, structured cohort discussions, stress-testing of real-world adoption and liability scenarios and review.
The collaborative also developed an AI Care Standard Evaluation Framework to support the adoption of the standard. The framework provides questions that will enable healthcare organizations to assess whether their patient-facing AI meets the standard.
In addition, the framework offers a structured review of how AI-generated patient communication is designed, governed and experienced by patients in real-world settings.
“AI is outpacing governance and oversight,” said Raj Ratwani, Ph.D., vice president of scientific affairs at MedStar Health Research Institute, in a press release. “The AI Care Standard provides the structure and rigor needed to evaluate and deploy AI responsibly in one of healthcare’s highest-impact, highest-risk domains: communication with patients.”
Raj is a co-chair of the PatientAI Collaborative, along with Bridget Duffy, M.D., former chief patient experience officer at Cleveland Clinic. Other members include leaders from Highmark Health, Stanford Medicine, HCA Healthcare, Hackensack Meridian Health, Patients for Patient Safety US, Vital and MATTER.
