Multiple Source Annotation of ECG T-Waves for Measuring QT in Drug Development

QT interval (adjusted for heart rate) of electrocardiogram (ECG) is the current measure for assessing cardiac safety of noncardiac drugs in drug development. It measures the length between the onset of the Q-wave and the offset of the T-wave. Many single-lead methods are developed to annotate the wave boundaries. While they agree quite closely on the onsets of the Q-waves, often times they differ by large margins on the offsets of the T-waves, since the T-waves are more variable. We propose three methods to combine the annotation results from multiple sources, which can either be annotations from different leads or annotations using different methods. The three methods are the meta-analysis methods for integrating independent and dependent sources and the Bayes-expectation-maximization (EM) algorithm method. The results from these information integrated methods are much better than those obtained from single-source methods, which is illustrated by a simulation study and real-data applications.