Construction of diagnostic prediction model for Wilson's disease.

Abstract

Wilson's disease, also known as hepatolenticular degeneration, is a rare human autosomal recessive inherited disorder of copper metabolism. The clinical manifestations are diverse, and the diagnosis and treatment are often delayed. The purpose of this study is to establish a new predictive diagnostic model of Wilson's disease and evaluate its predictive efficacy by multivariate regression analysis of small trauma, good accuracy, low cost, and quantifiable serological indicators, in order to identify Wilson's disease early, improve the diagnosis rate, and clarify the treatment plan. A retrospective analysis was performed on 127 patients with Wilson's disease admitted to the First People's Hospital of Yunnan Province from January 2003 to May 2022 as the experimental group and 73 patients with normal serological indicators who were not diagnosed with Wilson's disease. SPSS version 26.0 software was used for single factor screening and a multivariate binary logistic regression analysis to screen out independent factors. R version 4.1.0 software was used to establish an intuitive nomogram prediction model for the independent influencing factors included. The accuracy of the nomogram prediction model was evaluated and quantified by calculating the concordance index (C-index) and drawing the calibration curve. At the same time, the area under the curve (AUC) of the nomogram prediction model and the receiver operating characteristic (ROC) curve of the Leipzig score was calculated to compare the predictive ability of the nomogram model and the current Leipzig score for Wilson's disease. Alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (AKP), albumin (ALB), uric acid (UA), serum calcium (Ca), serum phosphorus (P), and hemoglobin (HGB) are closely related to the occurrence of Wilson's disease (p < 0.1). The prediction model of Wilson's disease contains seven independent predictors: ALT, AST, AKP, ALB, UA, Ca, and P. The AUC value of the prediction model was 0.971, and the C-index value was 0.972. The calibration curve was well fitted with the ideal curve. The nomogram prediction model had a good predictive effect on the occurrence of Wilson's disease; the ROC curve of Leipzig score was drawn, and the AUC value was calculated. The AUC of the Leipzig score was 0.969, indicating that the prediction model and the scoring system had predictive value, and the nomogram prediction model had a better predictive effect on the research objects of the center. ALT, AST, AKP, ALB, UA, Ca, and P are independent predictors of Wilson's disease, and can be used as early predictors. Based on the nomogram prediction model, the optimal threshold was determined to be 0.698, which was an important reference index for judging Wilson's disease. Compared with the Leipzig score, the nomogram prediction model has a relatively high sensitivity and specificity and has a good clinical application value.