National Academy of ⁢Medicine Issues AI Code of Conduct for Healthcare

Updated May 30, 2025
‍

The National ⁤Academy of Medicine (NAM) has published a comprehensive code of conduct to guide the ethical and effective use of artificial intelligence (AI) across the health sector. As⁤ AI tools become more prevalent in clinical, administrative, ⁣and research settings, the code aims to align health systems, developers, regulators, and patients around a shared set of values.

The ⁣AI Code of Conduct‍ for ⁤Health and Medicine⁤ focuses on safety,accountability,equity,and transparency while fostering innovation. The report emphasizes that managing risks thoughtfully is crucial to⁣ realizing AI’s potential in transforming medicine.

Built around ⁤six “Code Commitments”—Advance Humanity,Ensure ⁢Equity,Engage Impacted Individuals,Improve Workforce ⁢Well-Being,Monitor Performance,and Innovate and Learn—the framework serves ⁢as guiding principles for the development,deployment,and governance of health AI systems. Input was gathered from clinicians, ethicists, developers, researchers, and patient advocates from institutions including the Mayo Clinic and Google.

To put the Code ‍commitments into ⁣action,⁣ the⁢ report introduces a‍ “Tight-Loose-Tight” model of governance. This model ‍begins ‍with establishing⁤ shared ⁢vision and goals, ⁤followed by a⁤ phase where ⁤local organizations innovate ‍and adapt, and concludes ⁢with rigorous evaluation, ⁤transparency,⁤ and accountability. The NAM believes this approach allows for⁢ both centralized alignment and decentralized innovation, recognizing‍ the diversity of ⁢healthcare environments and the need for iterative learning.

The report also stresses the importance ⁢of standard-setting bodies and certification frameworks to assess adherence to these principles, encouraging public⁢ transparency to ⁢build trust among stakeholders. A key focus is ensuring AI ⁢systems ⁢do not worsen existing healthcare disparities. Standardized metrics should be used to identify and correct bias in data sets and model outputs, according to the NAM.

The report recommends targeted support and incentives to help‍ low-resource settings implement AI responsibly, as high-resource organizations may adopt AI tools more quickly. Developers are urged to create tools with⁤ built-in safeguards to minimize bias and maximize accessibility, and federal agencies are called upon to⁣ provide financial and regulatory incentives to encourage equitable AI deployment.

The code also addresses data privacy,⁢ transparency,⁢ and continuous monitoring of ⁢AI performance. Establishing quality and safety metrics will be essential ‍for evaluating AI’s impact on health outcomes. The report calls for⁣ a shared governance model that includes stakeholders from across the healthcare continuum, noting the absence of national⁤ standards for assessing AI tools.

The Code outlines specific responsibilities for developers, researchers, health systems, patients, ethicists, and federal agencies. ⁤Health systems are seen as critical actors in local ⁤adaptation and workforce ⁣training, with an opportunity to create financial incentives that⁢ support equitable and effective health⁣ AI.‍ They are ⁣also⁣ tasked with ensuring that implementation ‍promotes patient-centered care.

What’s next

The NAM urges stakeholders to adopt the six Code Commitments, ‍engage all stakeholders in AI governance, apply the Tight-Loose-Tight model, use standardized metrics, support⁣ equitable AI access,⁢ invest in workforce training, encourage federal support, and⁢ include ethicists and patient advocates in AI project planning.