Measuring Sleep in the Intensive Care Unit using Machine Learning

Schippers, Amber (2020) Measuring Sleep in the Intensive Care Unit using Machine Learning. Bachelor's Thesis, Artificial Intelligence.

Preview	Text AI_BA_2020_AmberSchippers.pdf Download (449kB) | Preview
	Text Toestemming.pdf Restricted to Registered users only Download (95kB)

Abstract

Sleep abnormalities occur frequently in Intensive Care Unit (ICU) patients, resulting in adverse effects on their health. It is important that their sleeping patterns are well understood in order to improve their sleep. Overnight electroencephalography (EEG) recordings are used to analyze sleeping patterns. The international norm is manual scoring by sleep experts following the American Academy of Sleep Medicine (AASM) criteria. When the severely disrupted EEG patterns of ICU patients were previously scored, low agreement between scorers was found. This suggests that the current standard for sleep analysis may not extend to ICU acquired data. Over a period of three years, 61 critically ill ICU patients have been monitored using EEG in the UMCG hospital in Groningen. Three machine learning algorithms (logistic regression, multi-class support vector machines and random forests) are trained on EEG recordings acquired from healthy subjects, and then tested on both the healthy recordings and EEG patterns acquired from ICU patients. The results show that the algorithms perform significantly worse on ICU subject data than they do on healthy subject data. This suggests that the current standard for sleep analysis is less suitable for the analysis of ICU patients' sleep.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Wiering, M.A.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	04 Aug 2020 13:21
Last Modified:	04 Aug 2020 13:21
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/22994

Actions (login required)

View Item