Trustworthy AI, digital twins and evaluation for cardiometabolic disease

We can offer:

Clinical prediction and risk modelling: CVD risk scores (SCORE2-Diabetes, QRISK3), DFU risk prediction, and multimodal modelling that combines imaging, ECG and EHR data. We are rigorous about methods, including detecting and handling temporal data leakage.

Digital twins and generative AI: patient-level digital twins for diabetes and CVD, synthetic data generation, and generative AI for wound assessment.

Trustworthy and equitable AI: explainability, sex- and gender-sensitive model development, bias auditing, and work on LLM safety and security, including prompt injection.

Evaluation and implementation science: mixed-methods evaluation of digital health systems, usability studies, user-needs research and readiness assessment for telemonitoring.

Data and governance: EHR and personal health record expertise, interoperability and data governance.

We have a strong network through BCS Faculty of Health and Care, EFMI, JMIR and BMJ HCI, plus links with NHS clinical partners. That puts us in a good position to lead evaluation, dissemination and exploitation work packages.

Looking for: consortia applying to 2027 HLTH calls on AI and digital tools for non-communicable diseases, digital twins, health data and equity in care.