Improving Automatic Smartwatch Electrocardiogram Diagnosis of Atrial Fibrillation by Identifying Regularity within Irregularity

Abstract

Smartwatches equipped with automatic atrial fibrillation (AF) detection through electrocardiogram (ECG) recording are increasingly prevalent. We have recently reported the limitations of the Apple Watch (AW) in correctly diagnosing AF. In this study, we aim to apply a data science approach to a large dataset of smartwatch ECGs in order to deliver an improved algorithm. We included 723 patients (579 patients for algorithm development and 144 patients for validation) who underwent ECG recording with an AW and a 12-lead ECG (21% had AF and 24% had no ECG abnormalities). Similar to the existing algorithm, we first screened for AF by detecting irregularities in ventricular intervals. However, as opposed to the existing algorithm, we included all ECGs (not applying quality or heart rate exclusion criteria) but we excluded ECGs in which we identified regular patterns within the irregular rhythms by screening for interval clusters. This "irregularly irregular" approach resulted in a significant improvement in accuracy compared to the existing AW algorithm (sensitivity of 90% versus 83%, specificity of 92% versus 79%, p < 0.01). Identifying regularity within irregular rhythms is an accurate yet inclusive method to detect AF using a smartwatch ECG.

Keywords: Apple Watch; algorithm; atrial fibrillation; electrocardiography; irregularity; mobile health; regularity; smartwatch; wearables.

Figure 1

The Lorenz plot of a patient with AF (a) and a patient with PACs (b). The diagonal line shows the line along which the perfectly regular intervals would be.

Figure 1

The Lorenz plot of a patient with AF (a) and a patient with PACs (b). The diagonal line shows the line along which the perfectly regular intervals would be.

Figure 2

Example of the Apple Watch ECG without abnormalities (a) and from an AF patient (b) [10].

Figure 3

The K-means clustering function in four different ECGs; (a) is a healthy ECG without any irregularity and clusters, there are also no clusters found; (b) is an ECG with AF with irregularity, but no clusters, which are also not found by the function; (c) is an ECG with PACs, so with irregularity and clusters are found; and (d) is an ECG with PVCs, so also with irregularity and clusters are found.

Figure 3

The K-means clustering function in four different ECGs; (a) is a healthy ECG without any irregularity and clusters, there are also no clusters found; (b) is an ECG with AF with irregularity, but no clusters, which are also not found by the function; (c) is an ECG with PACs, so with irregularity and clusters are found; and (d) is an ECG with PVCs, so also with irregularity and clusters are found.

Figure 4

The patient is diagnosed with AF by the AW algorithm (a). However, the novel algorithm shows by finding clusters in the Lorenz plot (b), regularity within the irregularity, and so the novel algorithm correctly identifies this patient with no AF.

Figure 5

The ECG and found peaks by the algorithm (a) of a patient with PVCs, where the clusters were not identified through the algorithm (b).