Decision Support System for Early Prediction of Sepsis

Sepsis, an unusual immune system response to an infection is a life-threatening condition. An early detection of such a condition can result in saving about 20% of patients from mortality. This paper illustrates a new approach to detecting sepsis which is based on not only the data collected from the existing cases but also contains the essence of the traditional definitions used by the doctors for sepsis detection. A rule-based clinical support system has been proposed which is an amalgamation of the Artificial Intelligence (AI) models and the standard definition for detecting sepsis worldwide. The paper presents an overview of the traditional decision-making process for correctly predicting sepsis at an early stage using the rule-based approach. The dataset used is a labeled dataset from Kaggle for sepsis prediction titled "Prediction of Sepsis" provided by PhysioNet Computing in Cardiology Challenge in 2019. The amalgamation of these two would surely enhance the prediction models used for sepsis detection and facilitate its early detection making it life-saving for the patients.