The MIT Abdul Latif Jameel Clinic for Machine Learning in Health (Jameel Clinic) held an AI and Health Regulatory Policy Conference on December 4, sparking debates among faculty, regulators from the US, EU, and Nigeria, and industry experts about the regulation of AI in health. The conference centered around the question of whether the “black box” decision-making process of a safe and effective AI model requires a complete explanation for FDA approval, akin to the processes of drug approval.
The conference embraced the uncertainties surrounding the ongoing development of machine learning and considered whether regulators can keep apace with these developments, ensuring potential harm reduction and competitiveness in innovation. The event was organized under the Chatham House Rule to allow anonymous, frank discussions amongst a carefully selected audience of 100 attendees.
Rather than creating hype around AI in health, the Jameel Clinic aimed to keep regulators informed of the most recent developments in AI and offered a platform for proposing changes or new approaches to regulatory frameworks for AI’s health use cases, particularly in clinical settings and drug development.
In the current health industry landscape, grappling with labor shortages, rising costs, high burnout, and resignation rates among health professionals, the relevance of AI in medicine has grown. Therefore, the suggestion was made to prioritize the deployment of clinical AI as an operational tool over its use in patient diagnosis and treatment.
One issue pointed out by an attendee was the widespread lack of education not just among developers and healthcare systems, but also among patients and regulators. The attendee underscored the need to involve medical doctors, who are usually the primary users of clinical AI tools, in regulatory discussions.
Data availability was another significant concern for AI researchers at the conference. Many faced challenges such as limited access due to intellectual property constraints or the lack of large, high-quality datasets. One speaker suggested that the FDA could alleviate this burden by facilitating data creation, providing developers more freedom to focus on AI advancement without worrying about underinvestment due to data uncertainty.
The Jameel Clinic’s event received praise for its facilitation of an open, productive environment for discussions on AI health regulation. Attendees suggested that similar events should continue to keep regulators up-to-date with AI developments.
The crux of any regulatory system is safety, as one attendee highlighted, driving the need for a generational mindset that starts with safety and then progresses downstream.
