A cancer patient enrolled in a clinical trial today might wait months before regulators ever see the safety data generated by their treatment. The FDA wants to change that, and it has started running live experiments to prove the concept works.
The agency disclosed in late April 2026 that two active proof-of-concept studies are now testing whether continuous data feeds from trial sites can replace the traditional cycle of batch collection and delayed review. On April 28, the FDA also filed a formal request for public input on a broader pilot program that would bring artificial intelligence into the design and oversight of early-phase clinical research.
Together, the moves represent the FDA’s most concrete steps yet toward modernizing how experimental drugs are monitored, particularly in oncology, where development timelines routinely stretch beyond a decade and cost billions of dollars.
Two live trials are testing the concept
The FDA’s Real-Time Clinical Trials initiative, known as RTCT, is built around two ongoing studies. AstraZeneca is running a Phase 2 trial called TRAVERSE (NCT05951959), which evaluates acalabrutinib combined with venetoclax and rituximab in patients with previously untreated mantle cell lymphoma. Amgen is conducting a Phase 1b study called STREAM-SCLC, focused on small cell lung cancer.
The AstraZeneca study has already crossed a technical milestone. According to the FDA’s own announcement, the agency received and validated real-time safety signals from the TRAVERSE trial through a technology partner called Paradigm Health. That validation confirmed the data pipeline works end to end: adverse events, dosing changes, and key outcome measures now appear on regulators’ dashboards as they are recorded at clinical sites, rather than after weeks of data cleaning and aggregation.
“For patients like the ones in our trial, every week of delay in getting safety data reviewed is a week they may not have,” said Dr. Michael Wang, a mantle cell lymphoma specialist at MD Anderson Cancer Center who has followed the RTCT initiative closely. “If this infrastructure proves reliable, it could fundamentally change how quickly we learn whether a new combination is helping or harming people.”
The Amgen trial is less far along in public disclosure. The FDA has identified STREAM-SCLC as a second proof-of-concept but has not detailed which technology platform supports it, whether real-time feeds have been validated, or how its monitoring workflow compares with the AstraZeneca setup. Running two distinct pilots across different sponsors, platforms, and cancer types is the point: the agency needs to know whether its approach can generalize, not just work in a single carefully managed integration.
A parallel push to bring AI into trial design
The real-time data feeds are only one piece. The RFI filed on April 28, assigned document number 2026-08281 and scheduled for Federal Register publication the following day, asks researchers, drug sponsors, and technology companies how AI tools could reshape early-phase trials more broadly. The notice sketches a potential pilot program in which algorithms would help refine patient eligibility criteria, dose-escalation decisions, and safety signal detection.
This AI-focused effort builds on a draft guidance the FDA released in January 2025, titled “Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products” (docket FDA-2024-D-4689). That document proposed a risk-based framework for judging whether AI model outputs are reliable enough to inform approval decisions, with emphasis on transparency in model development, validation on clinically relevant datasets, and clear documentation of how algorithmic analyses feed into human regulatory judgment.
The draft guidance remains exactly that: a draft. It is not binding, and its final form will depend on public comments that have been submitted but not yet summarized by the agency. Still, it provides the intellectual scaffolding for both the RTCT pilots and the new AI optimization RFI.
What the FDA has not yet answered
For all the forward motion, several critical gaps remain.
The agency has not published measurable benchmarks for what “faster” actually means. A typical Phase 1 oncology trial takes roughly six to seven years from first patient enrolled to regulatory submission, according to a 2023 analysis by the Tufts Center for the Study of Drug Development. The FDA has not estimated how many months real-time monitoring or AI-assisted design could shave off that timeline, making it difficult to judge whether the initiative will produce meaningful acceleration or remain a narrow technical exercise.
The Amgen study illustrates the transparency gap. Without a named technology partner, a confirmed validation milestone, or even a public registry ID for STREAM-SCLC, outside observers cannot independently assess how replicable the RTCT infrastructure is beyond the AstraZeneca pilot.
Questions also remain about how real-time feeds and AI-generated alerts will integrate with existing safety reporting systems. The FDA has pointed to the HHS safety reporting hub as part of the broader infrastructure, but has not detailed data standards, cybersecurity protocols, or how responsibility will be divided between sponsors’ monitoring teams and FDA reviewers when a machine flags a potential safety problem.
Perhaps most important, the agency has not defined where the line falls between AI as a support tool and AI as a decision-maker. The draft guidance frames algorithms as aids to human regulators, but the RFI hints at more ambitious applications, such as adaptive trial designs steered by machine learning. Without published guardrails, the distinction between “AI-assisted” and “AI-driven” remains blurry.
Why the timeline matters for patients
The stakes are not abstract. Mantle cell lymphoma, the cancer targeted in the TRAVERSE trial, has a median survival of roughly four to five years after diagnosis, according to the Leukemia and Lymphoma Society. Small cell lung cancer, the focus of Amgen’s study, is even more aggressive, with a five-year survival rate below 7 percent per American Cancer Society data. For patients with these diagnoses, shaving even a few months off the regulatory review cycle is not a bureaucratic nicety. It is the difference between accessing a potentially effective therapy and running out of time.
That urgency is also why scrutiny matters. Faster data flows are only valuable if they maintain the rigor that keeps unsafe drugs off the market. The FDA’s challenge is proving that real-time monitoring and AI tools can accelerate decisions without cutting corners on the safety signals that batch review was designed to catch.
Proof will come when the pilots produce results
The agency has taken verifiable, concrete steps: two live pilots, one validated data pipeline, and a formal call for public input on expanding AI’s role. What it has not yet delivered is evidence that these experiments will translate into shorter timelines, broader access, or better outcomes for the patients enrolled in trials right now. That proof will come only when the pilots produce results, and when the FDA shows it can scale what works beyond a pair of oncology studies.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
