Nursing assessment and decision tree based predictions of perioperative cardiac morbidity and mortality

Abstract

Morbidity and mortality remain high for patients with preexisting cardiac disease undergoing non-cardiac surgery. Critical care nurses can identify high-risk surgical patients preoperatively, through evaluation of their hemodynamic parameters, and intervene to optimize hemodynamic functions hence reduce patient morbidity and mortality. This study aimed at proving that the nursing assessment can predict perioperative cardiac morbidity and mortality, suggesting a system composed of several classifiers to predict morbidity and mortality for cardiac patients undergoing non-cardiac surgery, and validating Decision tree-based technique through comparing it with further two techniques (memory-based reasoning and neural network).The study included 461 cardiac patients classified acording to their outcomes into 3 groups, mortality group (83 patients), morbidity group (323 patients), and group III that passed operation without complications (55 patients).The results revealed a highly statistical significant difference among the three groups regarding all medical risks, cardiac assessment findings, and operative risks, and their intra and post operative outcomes. Decision tree based prediction model is a predictive model that, as its name implies, viewed as a tree. A decision tree is a flow-chart-like tree structure; it is a way of representing a series of rules that lead to a class or value. As a result of applying this method to a training set, a hierarchical structure of classifying rules is created and the outcomes can be predicted. This technique utilized to prove that the nursing assessed data could predict the mortality and morbidity of cardiac patient undergoing noncardiac surgery. The results prove accuracy of the technique of about 81.78% which concluded that the nursing assessment can predict perioperative cardiac morbidity and mortality in cardiac patients underrgoing non-cardiac surgery.