Predicting major organ complications in primary Sjögren's disease using a machine learning ensemble strategy: a dual-center retrospective clinical study.

Objectives: Primary Sjögren's disease (SjD) is a systemic autoimmune disease whose major organ complications severely affect patient prognosis. Currently, there is a lack of efficient and accurate tools for early identification of high-risk patients. Methods: This study retrospectively collected clinical data from 232 SjD patients, including demographic characteristics, clinical symptoms, and laboratory indicators. Three machine learning algorithms-logistic regression (LR), support vector machine (SVM), and random forest (RF)-were used to construct prediction models, and a soft voting ensemble strategy was applied to combine model outputs. Feature importance was analyzed using the SHAP framework. Results: The ensemble model performed best on the test set, with an AUC of 0.725, accuracy of 71%, and a negative predictive value (NPV) of 79%. SHAP analysis revealed that complement C3 and immunoglobulin G (IgG) were the most important predictors, with low C3 and high IgG levels significantly associated with complication risk. Conclusion: The ensemble model showed moderate discriminatory performance for identifying major organ complications in SjD using routinely available clinical variables. Complement C3 and IgG were identified as important predictors. The model may serve as a preliminary auxiliary risk-stratification tool, but further external validation and model optimization are required before clinical implementation.