The role of artificial neural networks in prediction of severe acute pancreatitis associated acute respiratory distress syndrome

Abstract

Background: Acute respiratory distress syndrome (ARDS) is a serious and common complication of severe acute pancreatitis (SAP), and the early identification and intervention of ARDS are particularly important. This study aimed to construct predictive models for ARDS using early indicators of inpatients. Methods: The 214 SAP patients included were randomly divided into training set and test set, and the back propagation artificial neural networks (BP ANNs) and logistic regression (LR) models were trained by the training set, and then the performance of the two models was examined by the test set. Results: The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy and area under receiver operating characteristic curves (AUC) value of BP ANNs were 68.0%, 87.5%, 77.3%, 81.4%, 80.0%, 0.853 ± 0.054 (95%CI: 0.749-0.958). The sensitivity, specificity, PPV, NPV, accuracy and AUC value of logistic regression were 48.7%, 85.3%, 65.5%, 74.4%, 72.0%, 0.799 ± 0.045 (95%CI: 0.710-0.888). There were no significant differences between the BP ANNs and LR models in these parameters. Conclusion: Compared to the LR model, the BP ANNs model showed only a mild improvement in predictive performance in our study without significant differences. However, we still believe BP ANNs models have a bright future in clinical outcome prediction.