Airflow recovery from thoracic and abdominal movements using synchrosqueezing transform and locally stationary Gaussian process regression

A wealth of information about respiratory system is encoded in the airflow signal. While direct measurement via spirometer with an occlusive seal gold standard, this may not be practical for ambulatory monitoring patients. Advances sensor technology have made motion thorax and abdomen feasible small inexpensive devices, but estimating from these time series challenging due to presence complicated nonstationary oscillatory signals. To properly extract relevant features thoracic abdominal movement, a nonlinear-type time-frequency analysis tool, synchrosqueezing transform, employed; are then used estimate by locally stationary Gaussian process regression. It shown that, using dataset that contains signals under normal sleep conditions, accurate out-of-sample predictions, hence precise estimation important physiological quantity, inspiration respiration ratio, can achieved fitting proposed model both intra- inter-subject setups. The method also applied more case, where subjects general anesthesia underwent transitions pressure support unassisted ventilation further demonstrate utility method.