Differentiating IgA nephropathy (IgAN) from membranous nephropathy (MN) often requires an invasive renal biopsy. This study aimed to develop and validate an explainable machine learning model for non-invasive discrimination between IgAN and MN using integrated clinical and ultrasound features. We enrolled 308 patients, randomly splitting them into training (n = 216) and test (n = 92) sets. Clinical and renal ultrasound features were collected. Fifteen machine learning algorithms were evaluated, including ensemble methods, traditional classifiers, and advanced techniques. Feature selection was performed with LASSO, and model interpretability was provided via SHAP analysis. Baseline characteristics were balanced between training and test sets (all P > 0.05). LASSO identified 17 key variables, with SHAP analysis highlighting albumin, age, and eGFR as top predictors. Among all models, Discriminant Analysis achieved the best performance, with a training AUC of 0.962 and test AUC of 0.968. It also attained 94.0% accuracy in training and 88.0% in testing, supported by well-calibrated probability curves. SHAP confirmed albumin (mean |SHAP|= 0.184) as the most influential feature, followed by age (0.161) and eGFR (0.105). Despite testing complex methods, Discriminant Analysis showed superior generalization and consistency. We successfully developed an explainable machine learning model that accurately discriminates between IgAN and MN using non-invasive clinical and ultrasound parameters. The model offers high discriminative performance and interpretable predictions, which may support clinical decision-making as proof-of-concept evidence; prospective external multi-center validation remains necessary before clinical deployment.
Gan et al. (Mon,) studied this question.