Multi-sensor signal processing methods for home monitoring of cardiovascular and respiratory diseases
Javaid, Abdul Qadir Qadir
MetadataShow full item record
Cardiovascular and respiratory diseases are leading contributors of health problems in the world. The existing home monitoring devices for cardio-respiratory health are obtrusive and incapable of measuring a broad range of physiological parameters. In this context, this research investigated signals from existing and new measurement modalities for estimation of mechanical parameters of physiological function for home monitoring of cardiovascular and respiratory health. Specifically, over-night data from an under-the-mattress impulse radio ultra-wide band (IR-UWB) radar combined with the signals from a microphone sensor were analyzed using machine learning algorithms to detect sleep apnea, a sleep related respiratory disorder caused by involuntary cessation of breathing during sleep. In parallel, for monitoring cardiovascular health, the ballistocardiogram (BCG) signal, a measure of reactionary forces of the body as the blood is ejected into the aorta and vessels, was analyzed using a variety of wearable and unobtrusive sensors. Algorithms were developed to assess the relationship of BCG with existing hemodynamic measurement modalities to increase the breadth of clinical parameters estimated from BCG. Data driven algorithms were designed for estimation of systolic time intervals from BCG signals during walking and in non-ideal postures. Finally, this dissertation demonstrated methods to differentiate between compensated and decompensated heart failure patients based on pre-ejection period changes after a six-minute walk test. These methods can potentially lead to automated wearable system that can predict decompensation beforehand, allowing physicians to intervene accordingly.