FDA to Fast-Track Approvals with AI

The Food and Drug Administration (FDA) is set to revolutionize its drug and device approval process by integrating artificial intelligence (AI). This initiative, detailed in a recent JAMA article, aims to “radically increase efficiency” and expedite the delivery of crucial treatments to patients.

Driven by a need to streamline operations following workforce reductions, the FDA sees AI as a pivotal tool. Other key priorities include a thorough review of potentially harmful ingredients in U.S. food and accelerating final approval stages to weeks, inspired by the rapid response during the COVID-19 pandemic.

“The FDA will focus on faster cures, meaningful treatments for neglected diseases, healthier food, and rebuilding public trust,” stated Dr. Marty Makary, the FDA Commissioner, and Dr. Vinay Prasad in their JAMA article. They emphasized the success of Operation Warp Speed as a model for accelerated approvals.

This push aligns with the Trump administration’s broader agenda to ease regulations for major industries. The agency has already started urging food manufacturers to eliminate artificial dyes, reflecting a commitment to public health.

However, the AI rollout has faced skepticism. Concerns have been raised about whether AI can truly handle the complexities of drug and device approval applications. Stephen Holland, a former advisor to the House Committee on Energy and Commerce, expressed cautious optimism but emphasized the need for concrete results.

The AI implementation follows a controversial report by the MAHA Commission, which included fabricated scientific research generated by AI. This raises questions about the reliability of AI in critical decision-making processes.

To further expedite approvals, the FDA is considering accepting a single major patient study instead of the traditional two, citing the pandemic as justification. “We believe this demonstrates that rapid reviews are possible,” Drs. Makary and Prasad asserted.

Critics like Mr. Holland point out that the pandemic response involved reassigning staff from routine duties to focus on COVID-19 products. Furthermore, the FDA’s workforce has shrunk recently, with nearly 2,000 employees leaving.

The FDA recently introduced Elsa, an AI large-language model, to assist in prioritizing inspections, summarizing side effects, and performing basic product-review tasks. While officials believe AI can enhance efficiency, current and former health officials note its limitations. Elsa’s character limit restricts its ability to perform complex data analysis, and its results require careful verification.

Some staff members have reported instances of “hallucination,” where Elsa generates false information. Dr. Makary clarified that the AI models are not trained on data submitted by the drug or medical device industry.

The FDA plans to scrutinize “chemically manipulated diets,” aiming to re-evaluate the benefit-to-harm balance of food additives. Despite proposed budget cuts, the food division is expected to receive additional funding.

Agency officials are navigating a complex landscape, balancing the need to partner with industry while avoiding undue influence. Dr. Reshma Ramachandran of Yale criticized a closed-door listening tour with drug industry CEOs, questioning the agency’s commitment to impartiality.

Christina Jewett covers the Food and Drug Administration, focusing on drugs, medical devices, food safety, and tobacco policy.