Network Analysis and Nomogram in the Novel Classification and Prognosis Prediction of Advanced Schistosomiasis Japonica

Abstract

ABSTRACT. Clinical classification of advanced schistosomiasis japonica is important for treatment options and prognosis prediction. Network analysis was used to solve the problem of complexity and co-occurrence complications in classification of advanced schistosomiasis. A total of 4,125 retrospective patients were enrolled and divided randomly into a training cohort (n = 2,888) and a validation cohort (n = 1,237). Network analysis was used to cluster the isolated complications of advanced schistosomiasis. The accuracy of the network was evaluated. Nomograms based on the clustered complications were built to predict 1- to 5-year survival rates in advanced schistosomiasis. The predictive performance of the nomogram was also evaluated and validated. Fifteen isolated complications were identified: metabolic syndromes, minimal hepatic encephalopathy, hepatic encephalopathy, chronic obstructive pulmonary disease, pulmonary hypertension, respiratory failure, right heart failure, gastroesophageal variceal bleeding, gastrointestinal ulcer bleeding, splenomegaly, fibrosis, chronic kidney disease, ascites, colorectal polyp, and colorectal cancer. Through network analysis, three major clustered complications were achieved—namely, schistosomal abnormal metabolic syndromes (related to chronic metabolic abnormalities), schistosomal abnormal hemodynamics syndromes (related to severe portal hypertension and portosystemic shunting), and schistosomal inflammatory granulomatous syndromes (related to granulomatous inflammation). The nomograms showed a good performance in prognosis prediction of advanced schistosomiasis. The novel classification-based nomogram was useful in predicting the survival rate in advanced schistosomiasis japonica.