Automated Detection of Sleep Disorder-Related Events from Polysomnographic Data

This work presents our effort in analyzing human bio signals collected during sleep studies, to automatically detect events related to sleep disorders. We experiment with real sleep data collected using standard Polysomnography (PSG), and we detect events of interest from EEG signals, by segmenting the signal, extracting descriptive features from each segment, and applying supervised learning for classification. Our preliminary experimental results show that the event detection goal can be successfully achieved, while our methods are general enough to be directly applied to sleep data collected using alternative, noninvasive sensors.