JMIR Publications

ABSTRACT

Background and Aims: Atrial fibrillation (AF), the most prevalent cardiac arrhythmia, affects 2-4% of the global adult population and is linked with an increased risk of stroke. Early diagnosis of AF and atrial flutter (AFL) is crucial due to their association with stroke risk and the challenge posed by their often asymptomatic and episodic nature. Traditional ECG interpretation, requiring significant expert input, can be challenging, especially with poor-quality ECGs. This study evaluates the performance of a deep neural network (DNN) model in detecting AF/AFL from a large, heterogeneous set of long-term ambulatory ECG recordings, including real-world clinical data collected over six months at a university hospital. The aim was to assess its effectiveness in a setting reflecting the diversity and complexity of real-world clinical data.

Methods:

The research combined public datasets totaling 10,896 patients and authentic long-term ECG recordings from 4,346 patients at Kuopio University Hospital for the DNN model’s development. Its clinical accuracy and generalizability were assessed using clinical data and 1039 anonymized long-term ECG recordings from 1010 patients, all thoroughly reviewed and annotated by experts.

Results:

The DNN model demonstrated high effectiveness, achieving 96.35% sensitivity and over 99.99% specificity in time-level AF/AFL detection. At the patient level, it identified AF/AFL with 100% sensitivity and 98.77% specificity, producing false positives in only 11 (1.23%) patients, of which nine had other AF/AFL arrhythmias. The model was highly effective across diverse patient groups, including various ages, comorbidities, other arrhythmias, and poor-quality ECGs.

Conclusions:

This study showcases the potential of the DNN model to significantly enhance the interpretation of long-term ambulatory ECG recordings, offering an efficient tool for AF/AFL detection, which could decrease manual labor and streamline patient management.