Artificial intelligence driven monitoring and prediction of true sleep health on CPAP machines to improve patient care and reduce the harm of obstructive sleep apnea

In this project we aim to evaluate the potential to effectively predict and prevent the occurrence of obstructive sleep apnea during sleep therapy using multiple sensor inputs in conventional CPAP machines. This will be based on experimental tests during both non-REM and REM sleep and standard polysomnography (PSG) measurements. This project is expected to assess the effectiveness of proactive management and treatment of OSA using real-time monitoring of air pressure and airflow and deep learning predictive models while bringing into account the sleep stages detected by EEG signals. While preliminary work has proven the effectiveness of predictions using airflow and air pressure sensors, bringing sleep staging via analysis of EEG signal and the effect of other additional sensor will be explored. Accurate prediction and prevention of these events with respect to sleep staging could improve clinical treatment of people suffering from obstructive sleep apnea (OSA) syndrome.

Scott Lowe
Faculty Supervisor: 
Kamal El-Sankary;Sageev Ooore
Nova Scotia
Partner University: