24 February 2026 | The Ray Dolby Centre, University of Cambridge
At CCAIM, our mission is to develop AI that transforms medicine — moving from academic insight to real-world clinical impact. That ambition was on full display on 24 February, when CCAIM co-hosted “Partnering with Cambridge: AI to revolutionise clinical trials” alongside Cambridge University Health Partners (CUHP) at the Ray Dolby Centre.
Over 70 senior decision-makers from more than 45 organisations joined the discussion, exactly the kind of cross-sector conversation needed to reimagine how clinical trials are designed, evaluated, and delivered.
Professor Mihaela van der Schaar, Director of CCAIM, opened the event with a video address outlining the depth and ambition of CCAIM’s clinical research programme.
She reflected on over a decade of work in causal inference and causality and how these ideas are now being operationalised into practical tools. She presented work on digital twins for trial simulation, adaptive trial designs that update protocols in real time as evidence accumulates, and AI for pharmacology that moves beyond pattern recognition toward mechanistic reasoning about how drugs behave in heterogeneous patient populations. You can read more about her research on clinical trials here.
Looking ahead, she highlighted the emerging role of AI agents as the next frontier in clinical development. a vision she develops in her recent white paper. She argues that the central challenge in modern trials is no longer simply data scarcity, but fragmented reasoning. Biological, statistical, operational, and regulatory risks are too often assessed in isolation. Her proposed answer is a coordinated multi-agent ML framework operating over shared causal digital twins – one that treats clinical trials as continuously learning systems, progressively de-risking development by reasoning across biological, statistical, operational, and regulatory evidence in concert rather than in isolation.
CCAIM in the working sessions: digital twins under the microscope
The first working session: Human-less trials? The role of AI-enabled digital twins, moved from vision to implementation.
Professor Eoin McKinney, CCAIM faculty, opened with a fundamental challenge: we often lack clarity about the true quality and completeness of clinical data. He presented AutoPrognosis framework, an automated machine learning system for building robust clinical prediction models, alongside approaches to synthetic data generation that introduce realistic patient heterogeneity into trial design.
He also introduced SyncTwin, method for constructing synthetic control arms – replacing or augmenting placebo groups with AI-generated counterparts matched to real patients.
Research Student Silas Ruhrberg Estévez, CCAIM collaborator and Summer School organiser, took a more critical lens, examining the limitations of current randomised controlled trials and the practical constraints of external synthetic control arms. He discussed Calm-DT, a context-adaptive digital twin framework designed to respond to the inherent complexity of clinical data and settings. He concluded by highlighting the need for larger, more diverse datasets and interdisciplinary collaboration with clinicians and pharmacologists to strengthen the biological validity of these approaches.
Together, their presentations made clear that the question is no longer whether digital twins can contribute to clinical trials, but how to build them with sufficient robustness, transparency, and validation to earn trust. The discussion that followed reflected strong engagement across academia, industry, and regulatory perspectives.
The session also featured a compelling use case from Professor Richard Gilbertson: a virtual child model for testing treatments for rare paediatric brain cancers, where recruitment is exceptionally difficult and the stakes are high. Professor Barbara Pierscionek provided essential ethical grounding, reminding the audience that digital twins do not eliminate responsibility. Bias in training data, environmental costs of AI infrastructure, and public explainability remain critical concerns.
Patients, data, and regulation
The afternoon widened the conversation beyond modelling to the broader ecosystem shaping AI-driven clinical trials.
Rory Cellan-Jones, award-winning broadcaster and Parkinson’s disease advocate, brought the patient perspective into focus, emphasising that digital twins and AI-enabled trials must be clearly explained if they are to earn public trust.
Dr Melanie Ivarsson OBE, the newly appointed CEO of the UK’s new Health Data Research Service, highlighted the importance of building secure, high-quality national data infrastructure to support clinical AI at scale. Without trusted and accessible health data, even the most sophisticated models cannot translate into practice.
Professor Alastair Denniston, Chair of the MHRA National Commission on the Regulation of AI in Healthcare, the body advising the UK medicines regulator on AI oversight, closed the formal programme with a case for regulation that enables responsible innovation while safeguarding patients.
Together, these contributions reinforced that advancing AI in clinical trials is not only a technical challenge, but one that depends equally on trust, infrastructure, and governance.
A starting point
As Lord James O’Shaughnessy noted in closing, this event marked a beginning. Transformative collaborations do not emerge from a single afternoon, but they do begin with shared ambition and serious dialogue.
It was a pleasure to welcome representatives from CCAIM’s industry partners including AstraZeneca, McKinsey & Company, and Sanofi, whose ongoing collaboration reflects a shared commitment to advancing rigorous, translational AI in clinical development.
If you’re interested in collaborating, we would be delighted to hear from you.