The AMA’s outlined principles emphasize the need for extensive policies aimed at mitigating risks to both patients and physicians when using AI technologies. These new principles, according to a Nov. 28 news release from the AMA, will help to maximize the benefits of AI in healthcare while minimizing potential harms.
Key concepts within the AMA principles include:
- Oversight: The AMA supports a whole-of-government approach to implementing governance policies addressing risks associated with healthcare AI.
- Transparency: The AMA advocates for legal mandates detailing key characteristics and information in the design, development and deployment processes of AI, including addressing potential sources of inequity.
- Disclosure and documentation: The principles call for adequate disclosure and documentation when AI directly affects patient care, access, medical decision-making, communications or medical records.
- Generative AI: To manage risks associated with generative AI, the AMA urges healthcare organizations to develop and adopt policies addressing potential negative impacts before its adoption and use.
- Privacy and security: Under these principles, AI developers are urged to design systems with privacy in mind, and organizations must implement safeguards to ensure responsible handling of personal information.
- Bias mitigation: The AMA said it advocates for proactive identification and mitigation of bias in AI algorithms to promote a fair, inclusive and discrimination-free healthcare system.
- Liability: AMA said it will continue to advocate for limitations on physician liability related to the use of AI-enabled technologies, aligning with existing legal approaches to medical liability.
Furthermore, the principles outlined by the AMA also encompass situations where payors utilize AI and algorithmic decision-making processes to establish coverage limits, make claim determinations, and shape benefit designs.
The AMA said when payors are using AI, they must ensure that the use of automated decision-making systems does not impede access to necessary care or systematically deny care to specific demographic groups.
