FDA Officials Outline AI and Real-World Evidence Integration Strategy at DIA Global Annual Meeting
核心洞察
FDA officials discussed strategies for integrating artificial intelligence with real-world evidence to advance drug development during a panel at the DIA Global Annual Meeting in Philadelphia.
The number of NDAs and BLAs incorporating RWE increased from four in FY 2023 to ten in FY 2025, while AI-related submissions have surpassed 1,000 across CDER and CBER (搜索).
AI excels at pattern recognition, patient identification, phenotyping, and signal detection, but is limited in identifying rare events, according to CDER senior policy advisor Hussein Ezzeldin.
PHILADELPHIA – Officials from the US Food and Drug Administration (FDA) outlined the agency's evolving approach to integrating artificial intelligence (AI) with real-world evidence (RWE) as complementary tools for drug development during a panel discussion at the DIA Global Annual Meeting on Tuesday.
"We really want to advance AI and RWE as complementary tools to advance drug development," said Marie Bradley, senior advisor for RWE in the Office of Medical Policy in the Center for Drug Evaluation and Research (CDER), who led the discussion with fellow FDA officials.
Growth in AI and RWE Submissions
The FDA has observed a notable increase in regulatory submissions incorporating these technologies. According to a recent report prepared under the Prescription Drug User Fee Act VII (PDUFA) agreement, the number of new drug applications (NDAs) and biologics license applications (BLAs) that incorporate RWE rose from four applications in fiscal year 2023 to ten in fiscal year 2025.
Bradley noted that on September 23, 2025, the FDA published a compilation of examples demonstrating how RWE has been utilized to support regulatory decisions in CDER and the Center for Biologics Evaluation and Research (CBER (搜索)) since 2011. An update published on June 3, 2026, expanded coverage to include the Center for Devices and Radiological Health (CDRH (搜索)), reflecting the breadth of RWE use across the agency's centers.
Anindita Saha, associate director for strategic initiatives at the Digital Health Center of Excellence within CDRH (搜索), reported that CDER and CBER (搜索) have received over 1,000 submissions incorporating AI elements – a dramatic increase from just one submission in 2016. The majority of these submissions are in oncology, followed by gastroenterology, neurology, and psychiatry.
"We are seeing a lot of AI in the nonclinical research and in the manufacturing and post marketing space," Saha said.
Capabilities and Limitations of AI
Hussein Ezzeldin, a senior policy advisor in the Office of Medical Policy at CDER, detailed the specific strengths of AI in drug development. "What is AI good at? It is good at finding patterns in the data. This is an area where you can see a lot of progress, you can go through a large amount of data and extract certain features. Patient identification is another area where you can expedite patient enrollment. Also signal identification can be another area," Ezzeldin said, adding that AI is also effective at phenotyping.
However, Ezzeldin cautioned about the technology's limitations. AI models "are really good at finding patterns; they are not good at finding rare events. Using AI to find a rare event, you may not get what you are asking for," he explained.
In clinical trials, AI is being utilized to establish endpoints, select patients, and predict outcomes, according to Saha.
Industry Feedback on Draft Guidance
The panel also addressed industry response to the FDA's draft guidance on the use of AI in regulatory decision-making for drugs and biologics, issued in January 2025.
"We heard loud and clear that industry wants practical examples and practical use cases," Saha said, summarizing the feedback received during the comment period.
When asked by an audience member whether the FDA plans to make public information on the more than 1,000 AI-related submissions received, Saha acknowledged that some information is confidential and proprietary, limiting what the agency can share. However, she added that "we are trying to figure out the best way for disseminating this information."
The discussion reflects the FDA's ongoing efforts to develop a regulatory framework that accommodates rapidly evolving technologies while maintaining scientific rigor in the drug approval process.