Health systems are rapidly integrating artificial intelligence (AI) tools, driven by the promise of reduced costs and relief for overburdened healthcare providers. However, implementing AI can pose significant challenges, particularly for organizations with limited resources. Clinicians often express concerns regarding the reliability and potential biases of AI outputs. Experts emphasize the importance of robust governance protocols to ensure that these models are applied appropriately and meet necessary performance standards. Unfortunately, many healthcare providers remain unaware of their organization’s AI policies.
In response, the Joint Commission has introduced a new voluntary certification program aimed at establishing guidelines for AI oversight. Launched earlier this month, this initiative seeks to acknowledge organizations that effectively manage data, mitigate risks, reduce biases, and maintain safety through ongoing evaluation and education.
William Walders, executive vice president and chief digital and information officer at the Joint Commission, stated, “The certification is designed to be accessible to organizations at various stages of governance maturity.” Walders, along with Ken Grubbs, the commission's chief nursing executive and vice president for accreditation and certification operations, shared insights about the certification and its implications for low-resource health systems during a discussion with Healthcare Dive.
During the conversation, Grubbs emphasized the importance of establishing a suitable governance framework that prioritizes patient safety while fostering workforce confidence in technology. He pointed out that effective data management is critical, highlighting the need for secure access and continuous monitoring of AI products within health systems. Organizations are encouraged to maintain a registry of these products, tracking any modifications and addressing quality and safety concerns.
Furthermore, Walders noted the inclusion of education and training within the certification. “It’s essential to provide role-specific training to ensure safe adoption of AI tools,” he explained. This approach aligns with how healthcare systems consistently prioritize safe practices among clinicians.
Regarding the program's impact, Walders observed substantial interest from health systems eager to navigate the challenges and complexities associated with AI adoption. He explained that the certification’s design aims to adapt to the rapid evolution of AI technology, making it flexible enough to suit different organization sizes and applications.
In response to inquiries about AI applications, Walders highlighted the need for organizations to evaluate each AI use case individually. For instance, while household items like washing machines may not require governance oversight, clinical tools such as CT scanners do warrant attention from governance structures.
As for ensuring that small, rural hospitals—often reliant on Medicaid—can adopt AI successfully, Grubbs mentioned that the certification was intentionally designed to be adaptable. Their goal was to present fundamental principles for safe AI use while allowing organizations to interpret the guidelines in a way that suits their unique circumstances. The feedback from the field review was integral in shaping the certification to avoid overwhelming requirements while emphasizing safety.
Walders recalled initial ideas that might have overly prescribed governance composition. They recognized that in smaller organizations, roles could overlap, making it impractical to dictate specific titles and responsibilities. Instead, the team aimed to create a framework that is realistic and accessible, ensuring that even clinics in less populated areas can effectively incorporate new AI technologies without bureaucratic complication.
