Risk prediction of postoperative renal dysfunction based on preoperative lipid profiles in renal transplant recipients: A retrospective cohort study

Background Renal dysfunction is a frequent complication after kidney transplantation, leading to poor prognosis and increased mortality. Abnormal blood lipids are closely related to renal dysfunction, yet their associations and mechanisms among renal transplant recipients remain unclear. This study aimed to establish an effective risk prediction model for renal dysfunction among RTRs based on abnormal lipid profiles using machine learning. Methods This retrospective cohort study recruited a cohort of 345 RTRs and followed up for one year after renal transplantation. Patients' demographic and clinical characteristics, including blood lipids, were retrieved from the electronic medical record system and analyzed using machine learning. Renal dysfunction was defined as estimated glomerular filtration rate (eGFR) < 60 mL/min /1.73 m ² . The cohort was randomly split into training (n = 276) and validation (n = 69) groups at a 4:1 ratio. Predictors of renal dysfunction were determined using three ML models: RandomForest, XGBoost, and LightGBM. Results During the one-year follow-up, 193 (55.9%) patients had renal dysfunction. A total of 20 demographic and clinical variables were selected to screen for significant predictors of renal dysfunction, and five were retained, including age, gender, HDL-C, non-HDL-C, and LDL-C, based on which a nomogram was developed. The nomogram showed good diagnostic performance with an area under the curve (AUC) of 0.87 in the training group and 0.81 in the validation group. Conclusions Our study showed that preoperative lipid profiles predicted postoperative renal function among RTRs, based on which we developed a risk prediction model. The model can quickly identify high-risk RTRs with renal dysfunction, which is crucial for optimizing patient management and improving the prognosis.