Artificial Intelligence Applications in Early Diagnosis of Sepsis Disease

Abstract

Sepsis is a major cause of death in intensive care units worldwide. Early diagnosis and treatment are crucial for improving patient survival and reducing organ dysfunction. Combining sepsis research and computer science advances creates predictive models for identifying patients at risk, enabling earlier intervention and better outcomes. The connected model, proposed one was used to evaluate machine learning algorithms across patient age cohorts (infant, elder, and all age) within the context of the study. The connected model, which is thought to consider the possibility of the patient's previous condition and/or conditions, in situations like illness that spreads over time, was compared with the non-connected model, which is thought to depend only on the current situation. The connected Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM), Convolution Neural Network (CNN), Random Forest (RF), and Extreme Gradient Boosting (XGBoost) machine learning models created for various patient cohorts improved the study's ability to predict sepsis in a shorter amount of time. According to analysis of proposed model, sepsis can be predicted in the infant patient cohort at 4th hour and in the elder and all age patient cohorts at 3rd hour