How is AI Transforming Clinical Trials Today?

The FDA has run clinical trials the same way for 60 years. Data flows from study sites to drug sponsors, who analyze it and then submit it to the agency.

The gap between when data is collected and when regulators see it can stretch to years. On April 28, 2026, the FDA announced it is changing that.

The agency unveiled two proof-of-concept real-time clinical trials in partnership with AstraZeneca and Amgen, in which AI extracts data directly from electronic health records and transmits it to both the FDA and the pharmaceutical company as it is collected.

The FDA has already received and validated signals from AstraZeneca's trial through Paradigm Health, the startup providing AI capabilities for the proof-of-concept work, establishing the technical feasibility of the framework.

"For 60 years, we've been conducting clinical trials in the same way, where key data signals can take years to reach the FDA," said FDA Commissioner Marty Makary. "We are boldly advancing a modern approach whereby FDA scientists can view safety signals and endpoints in real time as a trial progresses."

AstraZeneca is conducting a Phase 2 multi-site trial called TRAVERSE in patients with treatment-naive mantle cell lymphoma, with participation from MD Anderson Cancer Center and the University of Pennsylvania.

Amgen is conducting a Phase 1b trial called STREAM-SCLC in patients with limited-stage small cell lung carcinoma, with final site selection currently in process.

Both trials are testing the same core mechanism: AI replaces the manual process of pulling study data from electronic health records and entering it into a data-capture system, a process that has not changed in decades and accounts for a significant share of the 45% of drug development time that Makary attributed to paperwork and data entry.

Kent Thoelke, founder and CEO of Paradigm Health, described what makes the approach newly viable. Large language models now enable unstructured as well as structured data to be extracted and quality-controlled in real time in a way that was not previously possible.

What Changes If It Works

"The whole purpose here is to accelerate cures," said Jeremy Walsh, the FDA's Chief AI Officer. The real-time approach could compress the gap between trial phases by allowing the FDA to view data continuously rather than waiting for discrete phase submissions. At present, there is typically a hiatus between phases as data is compiled and reviewed.

Dr. Paul Burton, Chief Medical Officer at Amgen, noted that the current system has concentrated clinical trials at academic medical centers with long histories of running studies and entering data manually.

By enabling data to be pulled directly from electronic health records, trials can more easily be conducted at any center that maintains those records. "This program has the potential to be transformational on so many levels in how we do clinical research," he said.

For smaller biotech companies, the change could be particularly significant. Krish Ramadurai, a partner at AIX Ventures, described the data feedback loop as a make-or-break factor for startups.

A real-time system would allow them to shut down failing programs faster or fundraise earlier after getting a preliminary read on a promising drug. "Typically, the delay on that data-feedback loop makes or breaks a lot of companies," he said.

The FDA will accept comments on its RFI until May 29, 2026. Final pilot selection criteria are expected in July, with pilot selections complete in August.