NIHR funds 5-year programme to evaluate AI tool in clinical trial

The Artificial Intelligence to aid Decision making in Kidney Transplantation (AID-KT) programme, which has been awarded over £2.4m in NIHR Programme Grant for Applied Research, will explore how AI can help predict outcomes from kidney transplants, and how this information may help doctors and patients make decisions about their care. The tool will be tested in a prospective clinical trial to see if AI predictions can help to improve the use of organs offered for transplant, and their outcomes.

"Our work will not only develop models for clinical use, but will also test the models in a real-world clinical trial to assess clinical impact, usability and potential barriers to wider implementation"

The programme is led by Simon Knight, Associate Professor of Transplant Surgery (Nuffield Department of Surgical Sciences), who says, “There is currently a big gap in clinical AI research between model development and real-world clinical implementation. AI models have the potential to provide useful predictions that can aid the complex decision making around organ offers, but these tools must be validated, explainable and fair. Our work will not only develop models for clinical use, but will also test the models in a real-world clinical trial to assess clinical impact, usability and potential barriers to wider implementation.”

Associate Professor in AI for Digital Health Tingting Zhu (Institute of Biomedical Engineering, Department of Engineering Science) is Co-Investigator and Lead for Work Package 1, Model Development. Professor Zhu says, “The role of my team is to develop explainable AI models that mitigate bias and provide uncertainty quantification to ensure AI safety. We will facilitate the integration of these models into a clinical decision support tool for MHRA approval and clinical evaluation.”

The project builds on previous work undertaken in collaboration with the AI for Digital Health group at the University of Oxford, led by Professor Zhu. During this work, 20 years of UK Transplant Registry data were used to train explainable machine learning models that predict patient outcomes when organ offers are accepted or declined.

"Ultimately, this project has the potential to optimise organ utilisation and improve long-term survival outcomes for kidney transplant recipients across the UK."

Prof Zhu adds, “Securing this NIHR Programme Grant for Applied Research marks a pivotal transition for our work, moving us from retrospective theoretical modelling to real-world clinical deployment. This funding allows us to address the complex challenges of deploying AI in healthcare rigorously, such as quantifying uncertainty and mitigating bias to ensure fair and safe predictions. By transforming our explainable AI models into a regulated, clinically viable decision support tool, we hope to enhance the confidence and precision of transplant teams when evaluating organ offers. Ultimately, this project has the potential to optimise organ utilisation and improve long-term survival outcomes for kidney transplant recipients across the UK.”