Automatic arrhythmia detection based on heart beat interval series recorded through bed sensors during sleep

A high frequency of cardiac arrhythmias has been reported in sleep disordered patients. In order to detect the presence of arrhythmia during sleep, cardiac activity needs to be monitored. Several devices exist able to provide reliable Heart Rate Variability (HRV) measures in a minimally-intrusive way. Hence, there is the need for the development of robust methods for arrhythmia detection based on HRV measures. In the present study a method for automatic arrhythmia detection based on the analysis of an inter-beat series was developed and validated on recordings coming from the MIT-BIH Arrhythmia Database. The method was also applied on a beat to beat interval series obtained from the ballistocardiographic (BCG) signal of one subject, which was recorded during sleep using an innovative bed sensor which allowed for a noncontact and unobtrusive recording. When compared to the results of arrhythmia detection performed on the electrocardiographic (ECG) signal (concurrently acquired using polysomnography) using Cardioline Cube HOLTER analysis software, our method achieved a sensitivity of 55.6%, a specificity of 94.7% and an accuracy of 91.8%. Our results suggest that the bed sensor could represent a reliable tool to assist the clinician in arrhythmia detection.