Atrial fibrillation (AF) is the most frequently occurring form of arrhythmia, which induces multiple fatal diseases and impairs the quality of life in patients; thus, the study of the diagnostic methods for detecting AF is clinically important. Here we present a feature extraction method for the detection of AF using a ballistocardiogram (BCG), which is based on a physiological signal database collected by a non-contact sensor. The BCG signals, including both with AF and sinus rhythm (SR), were collected from 37 subjects during overnight sleep (approximately 8 hours). The signals were split into 2915 one-minute segments (AF: 1494, SR: 1421) without overlap and labeled as AF and SR. BCG signals were transformed into BCG energy signals in order to highlight the features of AF and SR BCG signals; and four new data sequences representing different characteristics of the BCG energy signals were generated. The mean value, variance, skewness, and kurtosis of the four data sequences were calculated and 16 features were extracted for each segment. Five machine learning algorithms were used for classification. The results of this study show that the support vector machine (SVM) performed the best among the five tested classifiers and achieved sensitivity, precision, and accuracy of 0.968, 0.928, and 0.945, respectively. These results indicate that the proposed feature extraction method can be well applied to AF and SR classification and may lay foundations for the development of systems for long-term home cardiac monitoring and AF screening.

Atrial fibrillation (AF) is the most common arrhythmia and can seriously threaten patient health. Research on AF detection carries important clinical significance. This manuscript proposes an AF detection method based on ballistocardiogram (BCG) signals collected by a noncontact sensor. We first constructed a BCG signal dataset consisting of 28,214 ten-second nonoverlapping segments collected from 45 inpatients during overnight sleep, including 9438 for AF, 9570 for sinus rhythm (SR), and 9206 for motion artifacts (MA). Then, we designed a residual convolutional neural network (CNN) for AF detection. The network has four modules, namely a downsampling convolutional module, a local feature learning module, a global feature learning module, and a classification module, and it extracts local and global features from BCG signals for AF detection. The model achieved precision, sensitivity, specificity, F1 score, and accuracy of 96.8%, 93.7%, 98.4%, 95.2%, and 96.8%, respectively. The results indicate that the AF detection method proposed in this manuscript could serve as a basis for long-term screening of AF at home based on BCG signal acquisition.