Predicting Adverse Outcomes in Kidney Transplant Recipients Using an Interpretable Model Based on Shear-Wave Elastography.

Introduction: Shear-wave elastography (SWE) is a promising noninvasive technique for measuring renal fibrosis after transplantation. This study aimed to develop an interpretable model to predict allograft deterioration in kidney transplant recipients and evaluate the predictive ability of SWE features. Methods: In this prospective cohort study, we performed SWE examinations on kidney transplant recipients at Renji Hospital between October 2020 and August 2023. The primary outcome was a composite of a 40% decline in estimated glomerular filtration rate or end-stage kidney disease. A total of 396 patients with stable renal allograft function were included. Five machine learning methods were used to construct predictive models. Results: Among all participants, 69 (17.4%) individuals reached the outcome. The XGBoost model with the addition of SWE features achieved the highest predictive performance with 20 repeats of nested tenfold cross-validation AUC of 0.870 (95% CI: 0.862-0.878) in the training dataset and 0.868 (95% CI: 0.801-0.935) in the validation dataset. Patients with higher medullary or cortical tissue stiffness had worse prognoses. A high level (>10 kPa) of medullary SWE was an independent risk predictor (adjusted OR, 2.68; 95% CI, 1.12-6.41). Conclusion: The joint use of SWE parameters and laboratory data significantly improved the risk prediction performance for a faster decline in allograft function. This interpretable XGBoost model may provide a readily available system to guide patient monitoring using noninvasive methods.