JMIR Publications

ABSTRACT

Background:

Wearable devices have evolved as screening tools for atrial fibrillation (AF). A photoplethysmographic (PPG) AF detection algorithm was developed and applied to a convenient smartphone-based device with good accuracy. However, patients with paroxysmal AF frequently exhibit premature atrial complexes (PACs), which result in poor unmanned AF detection, mainly because of rule-based or hand-crafted machine learning techniques limited in terms of diagnostic accuracy and reliability.

Objective:

We developed deep learning (DL) classifiers using PPG data to detect AF from the sinus rhythm (SR) in the presence of PACs after successful cardioversion.

Methods:

We examined 75 patients with AF who underwent successful elective direct-current cardioversion (DCC). Electrocardiogram and pulse oximetry data over a 15-minute period were obtained before and after DCC and labeled as AF or SR. A 1-dimensional convolutional neural network (1D-CNN) and recurrent neural network (RNN) were chosen as the two DL architectures. The PAC indicator estimated the burden of PACs on the PPG dataset. We defined a metric called the confidence level (CL) of AF or SR diagnosis and compared the CLs of true and false diagnoses. We also compared the diagnostic performance of 1D-CNN and RNN with previously developed AF detectors (support vector machine with root-mean square of successive difference of RR intervals + Shannon entropy, autocorrelation and ensemble by combining two prior methods) using a five-fold cross-validation process.

Results:

Among the 14,298 training samples containing PPG data, 7,157 samples were obtained during the post-DCC period. The PAC indicator estimated 2,132 out of 7,157 post-DCC samples (29.79%) had PACs. The diagnostic accuracy of AF vs. SR was 99.3% vs. 95.9% in 1D-CNN and 98.3% vs. 96.0% in RNN methods. The area under receiver operating characteristic curves of the two DL classifiers was 0.998 (95% confidence interval (CI) 0.995-1.000) for 1D-CNN and 0.996 (95% CI 0.993-0.998) for RNN, which were significantly higher than other AF detectors (P < .001). If we assumed that the dataset could emulate a sufficient number of patients in training, both DL classifiers could still correctly diagnose AF even when the PAC burden was >20% (91.1% and 91.5% for 1D-CNN and RNN, respectively). The average CLs for true vs. false classification were 98.6% vs. 80.5 % for 1D-CNN and 98.3% vs. 82.4% for RNN (P < .001 for all cases).

Conclusions:

New DL classifiers could detect AF using PPG monitoring signals with high diagnostic accuracy even with frequent PACs and could outperform previously developed AF detectors. Although diagnostic performance decreased as the burden of PACs increased, performance improved when samples from more patients were trained. Moreover, the reliability of the diagnosis could be indicated by the CL. Wearable devices sensing PPG signals and DL classifiers should be validated as tools to screen for AF.