Clinical Researcher—October 2026 (Volume 40, Issue 5)
TRIALS & TECHNOLOGIES
Pamela Tenaerts, MD; Ken Getz, MBA
Artificial intelligence (AI) is viewed as a potentially transformative force in the development of new therapies and has helped accelerate the discovery of new molecules. AI is also proving to accelerate the conduct of trials—and not a moment too soon, as the bottleneck in research shifts downstream away from discovery to in-human trials still hampered by operational friction.
Now, a new study from the Tufts Center for the Study of Drug Development (Tufts CSDD), conducted in collaboration with Medable, has taken an important step toward gaining critical understanding about how AI can help improve drug development.
AI enablement tools have been piloted across the industry for the past five years, but evidence of their economic value has been harder to establish. Companies often lack any kind of quantification of the actual return on investment (ROI) or the value proposition. Tufts CSDD and Medable were keen to do a study that would correlate hard numbers with a specific AI enablement solution.
Tufts’ analysis examined a clinical monitoring agent in oncology development and applied expected net present value (eNPV)—a widely accepted financial modeling tool customized to the clinical research industry—to estimate its economic impact. Notably, it is the first time that eNPV modeling based on actual use and benchmark data has been applied to quantify the value of an agentic AI solution deployed within a clinical trial as well as at the broader drug development-program level.
Measuring the Economics of Efficiency
The study’s analysis combined benchmarked oncology program and clinical trial data from Tufts CSDD with operational data from Medable, including time savings and agent costs. Tufts CSDD then applied the eNPV framework to quantify the financial value associated with those changes.
The analysis estimates eNPV gains of approximately $7.5 million for a Phase II oncology development program, $11.3 million for an oncology development program when the agentic AI solution was deployed across Phases II and III, and $21 million for a Phase III oncology development program. It also calculated eNPV-based return on investment multiples of 64x for deployment in a Phase II program and 82x for deployment in a Phase III development program.
Moreover, the model estimated direct operating cost reductions associated with onsite monitoring of approximately $4.4 million per Phase II oncology trial and $5.6 million per Phase III oncology trial. The analysis also identified administrative offsite monitoring efficiencies valued at approximately $600,000 for Phase II and $1.7 million for Phase III trials. Those savings, which represent clinical research associate (CRA) time that could be reallocated elsewhere, were not included in the eNPV calculations.
The direct operating savings reflect reductions in expenses, such as onsite monitoring and travel, while eNPV takes a broader view of how operational improvements and changes in development timelines affect the financial value of an oncology program.
Of note: the 64x and 82x ROI multiples reflect the difference between the net financial value generated by the agentic AI–enabled monitoring solution and the investment to deploy it. Savings from operating task efficiencies have not been included in the model since these efficiencies may largely result in reallocating CRA time elsewhere.
Why Time Matters So Much
Perhaps the most significant finding in the study revolves around something clinical development organizations put great value on and always seem to need more of: time.
The analysis estimated that agents could accelerate clinical development by approximately 18 weeks, with faster patient enrollment accounting for approximately 109 to 119 days, and earlier database lock saving another two weeks or so. Rather than agents directly recruiting patients, the modeled improvement reflects faster signal detection and more rapid response or remediation when enrollment issues arise. Essentially, efficiencies in one part of clinical operations have potentially great consequences on critical-path activities elsewhere on the trial.
Importantly, the underlying modeling did not rely on a single trial. Instead, Tufts CSDD combined real-world deployment data, oncology workload analyses, CRA interviews and validation, and established industry benchmarks to model potential effects on trial costs, timelines, and financial value.
This helps explain why the modeled economic impact is larger than the direct savings generated by automating individual tasks. Saving hours on administrative work compounds over time and across trials. Accelerated timelines impact not only direct monitoring costs but also the regulatory submission and market entry dates. Ultimately, time is a major driver of financial value.
Qualitative Measure: Refocusing Humans Where They Matter Most
Another potentially critical benefit that emerged from the study’s findings was the importance of how clinical research professionals spend their time.
CRAs are critical team members who perform work that requires keen judgment and strong relationship-building with investigative sites—tasks that are best handled by experienced professionals. However, their roles also tend to include a host of repetitive administrative duties such as reviewing information, documenting results, preparing for visits, identifying potential challenges, and coordinating follow-up.
A clinical monitoring agent can often assist with portions of that administrative work, leaving clinical judgment and decision-making with the CRA.
Sites may encounter issues, such as low enrollment, query resolution lag times, and other snags where CRAs can now become a strategic partner to the sites in finding solutions. Reducing the administrative burden around their work can allow these experienced professionals to devote more attention to such critical responsibilities.
AI can tackle some of the most tedious, mundane, and time-intensive activities so that CRAs can focus more directly on human relationships and more valuable responsibilities.
From Capability to Value
Much of the recent discussion around AI has focused on what it can do. Can AI perform a specific task correctly? Can it identify information faster? Can it reduce manual work? Increasingly, however, the more important question is whether those capabilities translate into measurable operational and economic value.
While the financial estimates in the Tufts study were specific to oncology indications, the model findings provide a starting point for extrapolating the data to other therapeutic areas and indications. Further, as sponsors increasingly ask whether AI agents can produce meaningful improvements in cost, time, quality, and the deployment of professional expertise, the framework used here offers a way to evaluate those investments against concrete operational and economic outcomes.
The shift from asking what AI can do to determining whether it creates demonstrable value represents an important evolution in the way the clinical research industry can expand AI usage strategically.
Pamela Tenaerts, MD, (pamela.tenaerts@medable.com) is Chief Medical Officer at Medable, where she aims to drive advancement of decentralized research methodologies with evidence-based best practices. She brings more than two decades of clinical research experience to her duties, having previously served as Executive Director of the Clinical Trials Transformation Initiative (CTTI), Director of Clinical Programs at Coaxia, and Director of the Clinical Research Center at Sarasota Memorial Hospital, in addition to other research positions.
Ken Getz, MBA, (Kenneth.getz@tufts.edu) is Executive Director and Professor at Tufts Center for the Study of Drug Development, Tufts University School of Medicine, where he studies pharmaceutical research and development management and execution; protocol design optimization; contract service provider and investigative site management; e-clinical technology and data usage; and patient and community engagement. He is also the Chairman of the Center for Information and Study on Clinical Research Participation (CISCRP), a nonprofit that he founded to educate and raise public and patient clinical research literacy, as well as the President of the Otsuka Patient Assistance Foundation.
