Comprehensive Analysis of Cardiogenic Vibrations for Automated Detection of Atrial Fibrillation Using Smartphone Mechanocardiograms

Abstract

Atrial fibrillation (AFib) is the most common sustained heart arrhythmia and is characterized by irregular and excessively frequent ventricular contractions. Early diagnosis of AFib is a key step in the prevention of stroke and heart failure. In this paper, we present a comprehensive time-frequency pattern analysis approach for automated detection of AFib from smartphone-derived seismocardiography (SCG) and gyrocardiography (GCG) signals. We sought to assess the diagnostic performance of a smartphone mechanocardiogram (MCG) by considering joint SCG-GCG recordings from 435 subjects including 190 AFib and 245 sinus rhythm cases. A fully automated AFib detection algorithm consisting of various signal processing and multidisciplinary feature engineering techniques was developed and evaluated through a large set of cross-validation (CV) data including 300 (AFib = 150) cardiac patients. The trained model was further tested on an unseen set of recordings including 135 (AFib = 40) subjects considered as cross-database (CD). The experimental results showed accuracy, sensitivity, and specificity of approximately 97%, 99%, and 95% for the CV study and up to 95%, 93%, and 97% for the CD test, respectively. The F 1 scores were 97% and 96% for the CV and CD, respectively. A positive predictive value of approximately 95% and 92% was obtained, respectively, for the validation and test sets suggesting high reproducibility and repeatability for mobile AFib detection. Moreover, the kappa coefficient of the method was 0.94 indicating a near-perfect agreement in rhythm classification between the smartphone algorithm and visual interpretation of telemetry recordings. The results support the feasibility of self-monitoring via easy-to-use and accessible MCGs.

Heart Rate Variability Analysis on Electrocardiograms, Seismocardiograms and Gyrocardiograms on Healthy Volunteers

“…Technological improvements and miniaturization of accelerometers make seismocardiography a useful non-invasive technique for examining cardiac activity [ 5 , 11 , 12 , 13 , 14 ]. The applications of SCG include heart monitoring during magnetic resonance imaging (MRI) scan [ 15 , 16 ], monitoring response to cardiac interventions [ 5 ] heart rate variability analysis [ 4 , 13 , 17 , 18 , 19 , 20 ], the detection of atrial fibrillation [ 6 , 21 , 22 ], heart failure [ 22 , 23 ] and the diagnosis of myocardial ischemia [ 5 , 24 , 25 , 26 ].…”

Gyrocardiography: A Review of the Definition, History, Waveform Description, and Applications

“…Gyrocardiography is defined in [ 22 ] as a local pulses signal, obtained by placing a gyroscope in the place of an accelerometer in seismocardiography [ 8 ]. The works published in 2019 on gyrocariography describe new topics and already known topics, such as the heart beat detection [ 9 ], detection of atrial fibrillation [ 64 , 65 ], hemodynamics analysis [ 42 , 66 , 67 ], pulse transit time measurement [ 68 ] and respiratory and cardiac gating [ 69 ]. The review of the state of seismocardiography by A. Taebi et al in 2019 reveals that the analysis of rotational vibrations may provide complementary information to the SCG signal analysis [ 70 ].…”

“…The GCG along with the SCG and BCG constitute the mechanocardiography (MCG) or vibrational cardiography (VCG). The first term was proposed by [ 24 , 49 , 55 , 58 , 60 , 64 , 65 , 71 , 73 , 77 , 88 ] and the second term was proposed by [ 9 , 73 , 79 , 80 ].…”

“…The detection of atrial fibrillation in gyrocardiography is based on the irregularities of inter-beat intervals [ 49 , 60 , 61 , 64 , 65 , 77 , 107 , 108 ]. In [ 108 ] the achieved average true positive rate (sensitivity) was 99.9% and an average true negative rate of 96.4% in leave-one-out cross-validation.…”

“…The classification was based on linear least-squares classifier. The method proposed in [ 64 ] is based on singular spectrum analysis, envelope detection, zero crossing rate, time domain, entropy and spectral features and classification using KSVM with a majority vote learner and LOOCV. The accuracy of this approach was 97%.…”