MORE | Fall 2025

Automated Detection of Sleep Apnea Using Machine Learning

Obstructive Sleep Apnea (OSA) is a serious, underdiagnosed condition because standard diagnostic tests are expensive and difficult to access. This research compared multiple machine learning (ML) algorithms for detecting OSA using non-invasive physiological data from pulse oximetry and heart rate variability. The research evaluated performance of the ML algorithms against gold standard data from clinic and simulated real-world constraints such as signal noise and reduced recording lengths. The findings demonstrate the potential for using ML with non-invasive physiological signals to detect OSA in real-world scenarios, which can be translated into wearable devices for scalable, at-home OSA screening.