Performance of a continuous real-time qt interval monitoring algorithm for the critical-care setting

We developed an automated continuous real-time QTc interval monitoring algorithm for the critical-care setting. The performance of the QT interval measurement algorithm was tested on the PhysioNet adult QT ECG dataset (n=105), and on a pediatric ECG dataset (n=20) and a neonatal dataset (n=24) recorded from intensive care units. The algorithm performance is measured by sensitivity (the ability to measure the QT interval), and accuracy (the difference between the automated QT measurements and cardiologists' manual annotations). We obtained 92% sensitivity in the adult group, 85% in the pediatric group and 75% in the neonatal group. On the 95 adult cases which had both an algorithm and a cardiologist measurement, the mean difference was 1 ms with a standard deviation of 35 ms. On the pediatric ECGs, the mean difference was -12 ms with a standard deviation of 20 ms. In the neonatal cases, the mean difference was -6 ms with a standard deviation of 12 ms.