Date of Graduation
12-2022
Document Type
Thesis
Degree Name
Master of Science in Industrial Engineering (MSIE)
Degree Level
Graduate
Department
Industrial Engineering
Advisor/Mentor
Shengfan Zhang
Committee Member
Xiao Liu
Second Committee Member
Eric Specking
Keywords
Infections, Mortality
Abstract
Sepsis is a host response to infection in both adults and children. It contributes to approximately 1.7 million cases annually with nearly 270,000 annual deaths in the United States. In the United States, the financial burden of sepsis on survivors and their families including the hospitals is over $4.8 billion, at approximately $64,280 per hospitalization. The first goal of this thesis research is to develop efficient machine learning models to predict pediatric sepsis accurately for inpatients. The second objective is to develop machine learning methods to determine how early sepsis can be detected to mitigate mortality. We examine data collected from the patient encounters at the Arkansas Children's Hospital between the May and October 2021. The raw data comprised two inpatient populations and one emergency patient population. This project focuses on the inpatient population classified as high-risk. Feature engineering, feature selection and extraction using Recursive Feature Elimination were performed on the data to build reliable and accurate models. Supervised learning techniques were implemented to predict pediatric sepsis.
Citation
Manson-Endeboh, G. (2022). Machine Learning for Early Detection of Pediatric Sepsis. Graduate Theses and Dissertations Retrieved from https://scholarworks.uark.edu/etd/4731