Research Paper Volume 13, Issue 6 pp 8817—8834

Accurate prediction of acute pancreatitis severity with integrative blood molecular measurements

(A) Dynamic changes of AUC scores in classifying training set AP cases with gradually decreased number of features. The 14 features identified in initial biomarker screening are: mean hemoglobin level, albumin-globulin ratio, alanine aminotransferase, thrombin time ratio, plateletcrit, albumin, prothrombin time, creatinine, serum potassium, BUN, prothrombin activity, mean hemoglobin concentration, triglyceride, lymph percentage. (B) ROC curve for the classification of AP training set; (C) Components of APSAVE and their corresponding coefficients; (D) ROC curve for the classification of AP test set; (E) Average ROC curve when training set and test set samples were reshuffled 100 times.

Figure 2. (A) Dynamic changes of AUC scores in classifying training set AP cases with gradually decreased number of features. The 14 features identified in initial biomarker screening are: mean hemoglobin level, albumin-globulin ratio, alanine aminotransferase, thrombin time ratio, plateletcrit, albumin, prothrombin time, creatinine, serum potassium, BUN, prothrombin activity, mean hemoglobin concentration, triglyceride, lymph percentage. (B) ROC curve for the classification of AP training set; (C) Components of APSAVE and their corresponding coefficients; (D) ROC curve for the classification of AP test set; (E) Average ROC curve when training set and test set samples were reshuffled 100 times.