Research

Our research sits at the interface of artificial intelligence, biomedical science and clinical practice. Every project is co-designed from the outset with an external partner alongside academic supervisors from across the University, so the work is built for translation into practice, not just publication.

Since 2024 we've recruited over 40 doctoral researchers across three cohorts, co-supervised across Informatics, Biological Sciences, Medicine & Veterinary Medicine, and Engineering, and co-designed with partners across industry, the NHS and other organisations. Their projects range from EEG biomarkers for rare genetic conditions to generative AI for drug design, from causal analysis of national health records to mechanistic models of antibiotic resistance. AI4BI builds on the track record of Edinburgh's predecessor Biomedical AI CDT, which trained over 120 doctoral researchers and produced 100+ peer-reviewed publications, including in Nature and Bioinformatics.

Research areas

The areas below reflect where our current projects sit. They're not a fixed or exclusive list, and we expect the mix to shift as new projects, students and partners join.

Our largest research cluster spans brain-computer interfaces, EEG biomarkers for conditions like epilepsy and autism, and data-driven approaches to understanding depression and other conditions whose symptoms and course vary enormously between individuals. The common thread is using AI to find structure in complex, noisy biological signals.

From ovarian and breast cancer to mesothelioma and brain tumours, our researchers are building AI tools that integrate imaging, tissue and clinical data to improve diagnosis and predict how individual patients will respond to treatment, including explainable models designed to earn clinicians' trust.

Cardiac MRI, retinal scans, brain imaging: our researchers are developing AI methods that extract more clinically useful information from medical images, often with less data or less supervision than conventional approaches require, and with an emphasis on knowing when a model's prediction should and shouldn't be trusted.

Only a fraction of the human genome codes for proteins, and most disease-causing variants sit outside it. Our researchers are building AI tools to interpret this "dark genome," mine multi-omic data at scale, and design new therapies directly, from generative models for drug and protein design to engineered enzyme replacement therapies for rare metabolic diseases.

Working with real-world NHS and population data, our researchers build tools that support clinical decisions in the moment (from post-surgical deterioration to intensive care) and reveal patterns in how care is delivered, from unscheduled care pathways to the effect of home energy retrofits on children's respiratory health.

Bacteria are evolving resistance to frontline antibiotics faster than new treatments can be developed. Current projects in this area combine mechanistic and statistical modelling with large-scale surveillance data to understand how resistance emerges and spreads, work that ranges from molecular mechanisms of resistance through to global antibiotic-use and resistance trends in priority pathogens.

How our research gets made

Every project is co-designed from the start by its academic supervisors and an external partner. Partners are engaged from project design through to results, helping shape objectives, reviewing progress, and identifying opportunities for real-world testing. Projects also include a placement that may be taken up with the partner organisation, giving everyone a genuine stake in translating the research into practice.