Correlation between SpO2 and Heart Rate in Sleep Apnea Detection

Sleep apnea is a potentially severe sleep disorder, which occurs when a person ’s breathing is disrupted during sleep. Polysomnography (PSG) is the standard approach for diagnosing sleep apnea. However, this test is quite obtrusive, time-consuming, and with limited access for patients in need of sleep apnea diagnosis. This diagnostic process may be simplified by applying machine learning techniques on data generated by wearable devices with pulse oximetry (SpO2) and heart rate sensors. Feature selection is an important step in the training of machine learning models, which reduces the number of features required to train the machine learning models and classify new query instances. Correlation between features is an indication of how one feature impacts another. In this paper, we study the correlation between SpO2 and heart rate in the detection of sleep apnea. We employ Pearson’s correlation and Spearman’s correlation to assess the linear relationship between these features, and Bland-Altman plots to evaluate the agreement between them. In order to validate the results, this study is conducted across 5 different datasets. From the methods employed, we Find that there is no correlation between SpO2 and heart rate, in sleep apnea. This lack of evidence of correlation necessitates the use of both features for training and both sensors for data collection, on a wearable device for sleep apnea detection.