Fusion-Based QT Interval Estimation for Enhanced Long QT Detection on a South African Population using Mobile ECG

Objective of the Study: This study compares QT interval estimation algorithms and their fusion on a South African population with a high prevalence of tuberculosis (TB) using a mobile ECG to evaluate the precision of QT interval measurement and long QT detection. Methods: The study evaluated one open-source signal processing-based QT-estimation algorithm and four commercial machine learning (ML)-based algorithms (from AliveCor Inc., PulseAI Ltd., GE Healthcare Technologies Inc (EK12), and Safebeat Rx Inc.). These five algorithms were combined using two fusion models, a median-based (MB) model and a linear regression-based (RB) model. These algorithms were validated on a South African Database (2050 2-lead ECGs) and tested on a separate database (692 patients with heart disease; 50% with long QT syndrome). Simultaneous expert overreads from 12-lead ECGs were used for reference. Major Results: The fusion algorithms demonstrated the highest classification performance and lowest errors in QT estimates. The mean difference showed 0.01±17.00 ms and-3.56±17.55 ms for the RB model and the MB model, respectively, on the validation set; and-7.69±21.17 ms (MB) and 0.14±21.91 ms (RB) on the test set. Conclusions: This work demonstrates that a fusion of five independently developed QT estimation methods significantly enhances the accuracy of automated QT analysis and long QT detection using a 2-lead mobile ECG. Significance to Biomedical Research: Accurate and timely detection of long QT is crucial for implementing appropriate medical interventions to mitigate the risk of lifethreatening arrhythmias. The fusion of multiple independent algorithms can improve performance and labels for future learning.