Jonsered

Foot ulcers are a common and serious complication of diabetes mellitus (DM) that can lead to infection, amputation, and higher mortality. This research aims to develop and validate prediction models, including machine learning-based and statistical models, to identify patients with diabetes at risk of developing foot ulcers. The study uses retrospective electronic health record data from primary care in the Västra Götaland Region (VGR) combined with demographic data from Statistics Sweden to improve risk identification by considering complex medical and socioeconomic factors. The study involves developing AI-based machine learning models trained on electronic health record data, including diagnostic codes, healthcare interventions, visit types, ECG parameters, and clinical notes. These models will be refined through cross-validation and uncertainty quantification techniques, then validated using a separate patient cohort. Statistical models will also be developed to explore causal relationships between risk factors and foot ulcer development. Both approaches will be compared for their strengths and weaknesses, with input from clinicians and patient representatives. Participants include adults aged 18 years or older with diabetes diagnoses or prescriptions for diabetes medication recorded between 2014 and mid-2025. Researchers will analyze patient data to identify predictors like neuropathy and previous ulcers, while validating model performance through measures such as sensitivity and positive predictive value. The study also focuses on transparency and clinical interpretability of the models, aiming to support early preventive care for high-risk individuals and reduce the burden of diabetic foot ulcers on patients and healthcare systems.