Sleep apnea (SA) is a common disorder involving the cessation of breathing during sleep. It can cause daytime hypersomnia, accidents, and, if allowed to progress, serious, chronic conditions like cardiovascular disease. Continuous positive airway pressure is an effective SA treatment. However, long waitlists impede timely diagnosis; overnight sleep studies involve well trained sleep physiologists scoring a polysomnograph (PSG). A PSG comprises multiple physiological signals, including multi-channel electroencephalography (EEG). Therefore, it is important to develop simplified and automated approaches to detect SA. This project combines principles of data wrangling, signal processing, and machine-learning. The ultimate aim is to develop a machine-learning model (examples: Fisher’s linear discriminant, support vector machine, random forrest classifier) that will help sleep physiologists identify periods of sleep apnea in overnight EEG recordings (but the data wrangling and signal processing will form a significant part of this project). You will use actual clinical data (courtesy of Dr Andy Veale at NZ Respiratory and Sleep Institute).
Ideally, you're a good Matlab programmer, with a keen interest in signal processing, and a willingness to grapple with some advanced concepts in machine-learning.
Starting point: Barnes, L. D., Lee, K., Kempa-Liehr, A. W., & Hallum, L. E. (2022). Detection of sleep apnea from single-channel electroencephalogram (EEG) using an explainable convolutional neural network (CNN). PLoS One, 17(9), e0272167.
Undergraduate
None
No lab has been assigned to this project