Modelling and Classification of Sepsis using Machine Learning

Sepsis is a serious condition which is usually fatal when untreated. It is necessary to treat the infection by considering the important vital signs along with the parameters in the ICU bed side monitoring. Nowadays large number of deaths occur due to sepsis. The current work aims to develop an automated machine learning algorithm to classify sepsis. The process involves extraction of features and usage of machine learning algorithms. The physionet/ CinC challenge 2019 dataset is used to test the methodology. The dataset consists of large number of records (40,336). Features are selected based on ID3 algorithm and entropy as the impurity measure. Machine learning classifiers such as RF, DT, LR, KNN, GNB, SVC (Linear, Polynomial, RBF) andANN are used for classification into normal and sepsis. The results indicate that Artificial Neural Network (ANN) provides the highest accuracy of 99.9821.