Researchers at the University of Surrey have developed an AI system that predicts a patient's knee X-ray appearance one year ahead, aiding in osteoarthritis management. The tool generates visual forecasts and risk scores, presented at MICCAI 2025. It promises faster, more transparent predictions for better patient care.
Osteoarthritis, affecting over 500 million people worldwide and the leading cause of disability in older adults, now has a new predictive tool. Developed by the University of Surrey's Centre for Vision, Speech and Signal Processing and Institute for People-Centred AI, the system uses nearly 50,000 knee X-rays from about 5,000 patients to forecast disease progression.
The AI, based on a diffusion model, creates realistic future X-rays and provides a personalized risk score. It identifies 16 key points in the joint to highlight potential changes, enhancing transparency for clinicians. This approach is nine times faster than similar tools, with greater efficiency and accuracy, potentially speeding integration into clinical practice.
David Butler, the lead author, explained: "We're used to medical AI tools that give a number or a prediction, but not much explanation. Our system not only predicts the likelihood of your knee getting worse -- it actually shows you a realistic image of what that future knee could look like. Seeing the two X-rays side by side -- one from today and one for next year -- is a powerful motivator. It helps doctors act sooner and gives patients a clearer picture of why sticking to their treatment plan or making lifestyle changes really matters. We think this can be a turning point in how we communicate risk and improve osteoarthritic knee care and other related conditions."
Gustavo Carneiro, Professor of AI and Machine Learning at Surrey, added: "Earlier AI systems could estimate the risk of osteoarthritis progression, but they were often slow, opaque and limited to numbers rather than clear images. Our approach takes a big step forward by generating realistic future X-rays quickly and by pinpointing the areas of the joint most likely to change. That extra visibility helps clinicians identify high-risk patients sooner and personalize their care in ways that were not previously practical."
The research, detailed in the journal Medical Image Computing and Computer Assisted Intervention (MICCAI 2025), suggests adaptations for conditions like lung or heart disease. The team seeks collaborations to implement it in hospitals.