Trials That Learn in Real Time: ARPA-H SURPASS, FDA RTCT, and the Future of Clinical Development
ARPA-H’s SURPASS program and the FDA’s real-time clinical trials pilot point to a future where trials learn continuously. Here’s what each one does and why it matters.
The Waiting Is The Problem
A new medicine can take more than a decade and up to $2 billion to develop, and roughly nine in ten candidates still fail. Much of that time is not science at all. It is waiting: for the next phase to start, for data to be entered, cleaned and locked, for a report to reach the people who make decisions.
Former FDA Commissioner Marty Makary has estimated that about 45% of drug development time is this kind of dead time. In 2026, two federal efforts set out to remove it. ARPA-H launched SURPASS to rebuild how trials are designed and run, and the FDA began piloting real-time clinical trials (RTCT) that let reviewers see agreed signals as a study unfolds.
Together they describe a future in which a trial learns continuously instead of reporting in batches. At OmniScience, that is the future we have been building toward.
What ARPA-H Is, And Why It Works Differently
The Advanced Research Projects Agency for Health (ARPA-H) is the health counterpart of the model that gave us the Defense Advanced Research Projects Agency (DARPA). Its stated mission is to deliver health breakthroughs in years, not decades, by funding work that traditional research grants or commercial markets will not readily take on.
Three features set it apart:
- Program managers with a thesis. Each program starts with one person's bold idea. Program managers serve limited terms, so fresh thinking keeps arriving.
- Milestones that gate funding. Every milestone unlocks the next tranche. Programs are built to fail fast or scale fast, and ARPA-H openly accepts productive failure as the price of transformative results.
- Teams across disciplines. ARPA-H assembles statisticians, engineers, clinicians, regulators and companies around a single problem, and absorbs technical and regulatory risk no single institution would carry alone.
That structure makes ARPA-H a distinctive player in the AI research landscape. Much AI-for-health work optimizes one task: a model that reads an image or predicts an outcome. ARPA-H programs aim to change whole systems: SURPASS applies this ambition to the clinical trial itself.
SURPASS: A Trial With No Dead Time
On September 30, 2026, HHS and ARPA-H launched SURPASS (Simulation-augmented, Real-time Platform Adaptive Seamless Trials). Its goal is to cut clinical development from more than a decade to under four years, with lower cost and fewer participants.
The idea is a perpetual, adaptive, phaseless platform trial. Instead of separate phases, each with its own startup, the program envisions shared infrastructure, common control groups, predictive models and real-time analysis that let a trial keep running as new treatments enter and leave. SURPASS pursues three coordinated technical areas:
- A phaseless design engine that brings digital twins and other predictive models into trial design, simulating clinical and operational outcomes before launch.
- A continuous inference engine that supports always-valid analysis on demand, so teams can adapt quickly while still producing rigorous evidence with smaller control groups.
- An agentic operations layer that automates startup and operations, onboards new treatment arms, and speeds data collection, cleaning and dataset construction.
We applaud this work by ARPA-H to deliver life-saving treatments to patients faster and more efficiently. SURPASS points toward a future where trials learn and adapt continuously from first in human to market approval, and we are excited to see that future take shape.
FDA Real-Time Clinical Trials: The Regulator Joins The Loop
In April 2026, the FDA announced its real-time clinical trials (RTCT) initiative. Under RTCT, reviewers can see safety and efficacy signals agreed in advance while a study is still running, rather than waiting for data to be collected, cleaned and submitted.
Two proof-of-concept oncology trials, sponsored by AstraZeneca and Amgen, are already reporting endpoints and signals to the agency as they accrue. The FDA also requested public input on scaling the pilot and on how AI can improve the speed and quality of decisions in early-phase trials. Its stated long-term aim is continuous trials across every phase of development.
RTCT does not create a new approval pathway. What it changes is timing. Data that once moved from site to sponsor to regulator over months can now be visible in days. That only works if the data underneath are trustworthy, traceable and current.
The RTCT concept gives regulators a direct, near-real-time view of live trials. For the SURPASS adaptive phaseless designs, this gives regulators a view into decisions and trial adaptations driven by continuous inference.
How OmniScience Powers the Trials of Tomorrow
OmniScience is on a mission to transform clinical trial operations to deliver life-changing medicines to patients faster. We are a trusted AI partner to leading life sciences organizations, advancing clinical R&D through deep expertise in clinical data science. We believe the future of clinical development depends on unifying fragmented trial data and transforming it into real-time intelligence so teams can move faster, decide with confidence, and deliver better outcomes for patients.
When we read the goals of SURPASS and RTCT, we recognized our own roadmap. Shared infrastructure, continuous inference, agentic operations, and real-time regulatory visibility all depend on the same thing: a living governed data foundation that people and regulators can trust.
We are delighted to be joining forces with some of the top minds in statistics, AI, clinical operations, and regulatory science who are working on these problems. No single company will build the future of clinical development alone. It will come from teams willing to share infrastructure, standards, and hard-won lessons, and we are proud to contribute ours.
The Future We Need
For decades, clinical trials have been run like a relay with long pauses between legs. SURPASS and RTCT point toward something closer to a single continuous run, where evidence builds every day and decisions follow the data, not the calendar.
The beneficiaries are patients who wait less, sponsors who learn sooner what works and what does not, and regulators who see the evidence as it forms. This is the direction the field needs, and we are excited to help build it.
Connect with us
Our mission is to transform clinical trial operations to deliver life-changing medicines to patients faster. We built Vivo, the first agentic AI-powered control tower for clinical trials.
Vivo unifies fragmented clinical, safety, and operational data, and delivers real-time, explainable insights, putting teams ahead of the trial, not behind it. To learn more about Vivo, contact us at hello@omniscience.bio or connect with us on LinkedIn.

