Sleep Apnea Prediction in the Post-Stroke Patient based on the Sleep, Pain and Depression Parameters

Current practices in the rehabilitation program of post-stroke patients do not include monitoring or assessment of sleep disorder, pain and depression measures, although they significantly affect motor and cognitive function for recovery. The objective of this study is to apply a mathematical model of multiple logistic regression to predict the severity of sleep apnea from blood oxygen saturation, pain and depression measures. Linear (min and max) and non-linear features (approximate entropy) from SpO2 signals combined with pain score, BMI score and age are predictive parameters to detect the severity of sleep apnea. The outcome of this research is believed to complement current rehabilitation intervention, particularly in assessing sleep apnea which further may facilitate early recovery of post-stroke patients.