The study developed and validated a machine learning model to predict hypersplenism in patients with Wilson's disease, a rare disorder of copper metabolism associated with cirrhosis. They enrolled 524 patients from Anhui Hospital from December 2019 to February 2025, of which 242 had hypersplenism. Using LASSO variable selection and SHAP analysis, they identified five key factors: WBC, PLT, A/G, CIV and PIIINP, with PIIINP being the most significant. The group with hypersplenism had significantly lower WBC, PLT and ceruloplasmin than the group without it (p<0.05). A/G, CIV and PIIINP were independent risk factors, while WBC and PLT were protective. The best SVM model achieved an AUC of 0.867 (95% CI: 0.830–0.904), an accuracy of 0.807 and an AUC of 0.771 (95% CI: 0.699–0.844) on the test set. The model shows good calibration (Brier score 0.146/0.206) and clinical utility according to DCA.