Machine learning models for early detection of urinary tract infections in kidney transplant patients.

BACKGROUND: Renal transplantation is the preferred treatment for end-stage chronic kidney disease but requires lifelong immunosuppression, increasing the risk of infections such as urinary tract infection (UTI). UTI in kidney transplant recipients can lead to serious complications, including acute kidney injury, reduced graft survival, and increased mortality. Machine learning can enhance risk detection accuracy and support proactive management of complications. This study aims to develop a machine learning-based approach for predicting urinary tract infections in kidney transplant recipients, facilitating the early identification of high-risk individuals. METHODS: Retrospective cohort of 29,340 anonymized records from a Colombian transplant center to develop and compare machine learning models for predicting UTI in post-kidney transplant patients. Fifteen machine learning models were evaluated using 41 sociodemographic and clinical variables. Model performance was assessed using metrics such as accuracy and area under the ROC curve (AUC). RESULTS: Both models are good, with Random Forest achieving 93.57% accuracy, 94.43% AUC, 92.5% sensitivity, and 95.2% specificity. XGBoost achieved 91.14% accuracy, 91.65% AUC, 90.8% sensitivity, and 92.6% specificity. Feature importance analysis revealed that key predictors included Body mass index (BMI), use of mofetil mycophenolate or sodium mycophenolate, donor type, hypertension, diabetes mellitus, use of steroids, re-transplantation, serum levels of tacrolimus and sex of the recipient. CONCLUSION: This study demonstrates the potential of machine learning models to predict UTI in kidney transplant patients, offering a tool for early intervention and personalized care. Implementing these models in clinical settings could improve patient outcomes and reduce healthcare costs by preventing severe complications associated.