Project Abstract

Sepsis is a life-threatening and time critical condition that requires early intervention to mitigate its high mortality rates, which escalate by 3% with every hour of delay in diagnosis. Machine learning offers a promising avenue for improving detection, thus improving patient outcomes. This research explores the application of machine learning models in predicting Sepsis six hours prior to onset symptoms using Neural Networks, Support Vector Machines (SVM) and Extreme Gradient Boosted Trees (XGBoost). Through a comparative analysis of these models, this research aims to discern the most accurate approach, primarily utilizing sepsis labels (1 for Sepsis, 0 for non-Sepsis) for assessment. Despite encountering a significant dataset imbalance, with a 172% disparity between sepsis-positive and sepsis-negative cases, Neural Networks emerged as the standout performer, demonstrating increased accuracy and efficiency. While XGBoost also exhibited promise in sepsis prediction, albeit with slightly lower accuracy compared to SVM and Neural Networks. Due to prolonged processing times, XGBoost was the only model that was used to predict Sepsis six hours prior to onset symptoms, yielding expected performance outcomes given the datasets characteristics. Despite imperfect results, these findings underscore the potential of machine learning in enhancing sepsis detection and ultimately reducing mortality rates associated with the condition. By highlighting the effectiveness of machine learning models in predicting sepsis onset, this research contributes to advancing healthcare practices and emphasises the importance of early intervention in managing sepsis. Moreover, this study endeavours to raise awareness about Sepsis, fostering greater understanding and potentially saving lives through informed recognition and proactive intervention strategies.

Keywords: Machine Learning, Computational Medicine, Data Science

 Conference Details

Session: Poster Session B at Poster Stand 34

Location: Sir Stanley Clarke Auditorium at Wednesday 8th 09:00 – 12:30

Markers: Xianghua Xie, Mark Jones

Course: BSc Computer Science, 3rd Year

Future Plans: I’m undecided