Aten described external control arms as a way to use data from outside a clinical trial to reduce the need for traditional control-group recruitment. Digital twins, meanwhile, could model how enrolled patients might progress through a study, helping researchers account for differences among participants and improve statistical analysis.

However, Aten cautioned that these approaches are only as reliable as the data behind them. “If the underlying data that you're using still has systematic confounding and bias, that could be problematic,” he said, noting that randomization remains an important tool for reducing bias in research.

Corso said regulators are increasingly providing guidance on AI in drug development, but questions remain about data quality, consistency and traceability. Researchers need to demonstrate how data was collected, how outcomes were measured and whether information can be traced back to its source.

“There are a lot of ways to take a blood pressure in clinical research,” Corso said, emphasizing that measurement consistency matters when real-world data supports research findings.

He also pointed to the continued reliance on paper-based processes at clinical research sites. Citing a Tufts study, Corso said 67 percent of informed consents in the United States are still completed on paper. Some consent documents can run dozens of pages, while sites continue to rely on spreadsheets and disconnected systems to manage trials.

“What we need is less paper,” Corso said. “And then systems that talk to each other.”

Franchetti, a nurse with decades of clinical research experience and a parent and caregiver of children with rare diseases, argued that technology cannot improve clinical trials if patients are not meaningfully involved in decisions.

She said patients should be consulted early about trial design, data use, and the potential role of AI. Their perspectives may differ by disease and treatment setting, making broad assumptions about patient acceptance risky.

One concern she hears increasingly is whether AI will replace the patient voice just as patients are beginning to gain greater influence in research.

“Am I going to be replaced?” Franchetti said, describing a question patients are asking as AI becomes more prominent in clinical research.

She called for patient involvement to become a standard part of trial development rather than an optional practice.

“All of these decisions are being made about patients for patients, but they're not being made with patients,” she said.

Franchetti said involving patients and research sites earlier can help identify burdensome procedures, clarify which outcomes matter to patients, and reduce the risk of participants dropping out. She also stressed the importance of ongoing feedback from people who reflect the population a trial is intended to serve.

The panelists also discussed the need for better data infrastructure to support AI-enabled trials and a more connected healthcare system.

Aten said electronic health records have created an important digital foundation, but more work is needed to make data accessible, reliable and reusable across research and care delivery. He pointed to initiatives involving the FDA, the Office of the National Coordinator for Health Information Technology and CMS as potential contributors to a more connected data environment.

He said the goal should be a system in which information can move with patients and be used when needed, while maintaining trust and reliability.

Corso identified inconsistent data standards across clinical research as a major obstacle. Unlike healthcare, which has made progress toward common data standards, life sciences research uses multiple standards that do not always work together. This makes it difficult to move information between systems and across regulatory submissions.

The discussion comes amid growing federal interest in modernizing clinical trials, including HHS's Operation Trialblazer and ARPA-H's SURPASS program. Aten said these efforts, along with proposed legislation to embed clinical trials into care delivery, could help accelerate changes the industry has discussed for years.

He emphasized that the opportunity extends beyond major academic medical centers to rural communities and smaller clinics that have historically had fewer opportunities to participate in research.

Corso described the current environment as a promising period for clinical research innovation, while cautioning that improvements to regulatory submissions alone will not solve the underlying operational challenges of conducting trials.