A Clinical Assessment Of Machine Learning Methods With Adaptive Synthetic Sampling Approach For Imbalanced Learning On Sepsis Prediction

Abstract

Severe health problems such as sepsis, which commences with the body fighting an infection, can result in septic shock, which drastically reduces blood pressure and results in organ failure. Infections in humans can be caused by specific bacteria, viruses, and fungus. Whenever the infection is sufficiently severe, our immune system might begin an attack, which could worsen and result in sepsis. Sepsis cannot be identified with a single test, making the diagnosis extremely challenging. Consequently, sepsis can be identified by several tests, including those that evaluate for infections, very low blood pressure, and an irregular heartbeat. To recognize sepsis in the healthcare sector nowadays, machine learning algorithms will be necessary. In addition to integrating patient data, machine learning algorithms can handle extremely complicated and important data. A multitude of machine learning methods are employed to detect sepsis. Analyzing the algorithms used to predict sepsis is the aim of this study analysis, the techniques for evaluating these machine learning algorithms' performance, as well as their drawbacks and restrictions. This work aims to provide a clear explanation of the significance of earlier research applied to sepsis prediction, as well as a clear understanding of machine learning methods for sepsis prediction for beginners.