Research & Development

Our research programme focuses on the methodological foundations of clinical AI evaluation — generating evidence that supports both policy development and the direct assessment of AI health tools.

Algorithmic Performance Across Populations

We investigate how AI diagnostic and predictive tools perform when evaluated on patient populations that differ by geography, ethnicity, socioeconomic status, and comorbidity profile — and we publish what we find, including null results.

Clinical Trial Design for AI Interventions

We develop and refine methodological frameworks for prospective clinical trials of AI health tools, addressing the particular statistical and ethical challenges that arise when the intervention is an adaptive algorithm.

Data Quality & Representativeness

We study the structure and limitations of health data used to train and validate clinical AI systems, with particular attention to data provenance, labelling practices, and the conditions under which training-distribution shift occurs in clinical deployment.

Collaboration

We collaborate with academic medical centres, independent research institutes, and clinical epidemiology groups. If your work intersects with clinical AI evaluation methodology, get in touch.