Jonsered

Researchers are studying diabetes mellitus DM to improve the prediction of foot ulcers, a serious and common complication in people with diabetes that can lead to infection, amputation, and higher mortality. The study aims to develop and validate prediction models using both artificial intelligence AI and traditional statistical methods, considering clinical and socioeconomic factors that current models may overlook. This research addresses the increasing global burden of diabetes and the need for better early identification of patients at risk to support preventive care and reduce healthcare costs. The study uses retrospective electronic health record data from primary care in the Vstra Gtaland Region VGR and demographic data from Statistics Sweden. Machine learning models will be developed and trained with cross-validation to predict the risk of foot ulcers, with uncertainty measured by conformal prediction. Statistical models will analyze causal relationships between risk factors and ulcer development. The models will be compared for strengths, weaknesses, and clinical interpretability, collaborating with patient representatives and clinicians. Participants include adult patients 18 years or older with diabetes diagnoses or prescriptions recorded between 2014 and June 2025. Data such as diagnostic codes, procedure codes, visit details, ECG parameters, and clinical notes will be used to identify risk predictors. The study will measure model performance by sensitivity, specificity, and positive predictive value PPV. Validation on an independent dataset will ensure robustness. The study runs through 2027, focusing on developing transparent, clinically useful prediction tools to improve preventive care for diabetic foot ulcers.