Automated QT interval analysis with confidence measures

Current automated methods for QT interval analysis suffer from the absence of a confidence value in the resulting measurements. In this paper we present a new approach to QT interval analysis, which produces both a segmentation of the ECG together with an associated measure of confidence in the segmentation. The method is based on a novel hidden Markov model architecture, which is designed to prevent unrealistic segmentations. We utilise the probabilistic nature of the model to derive a confidence measure based upon the log likelihood of the ECG waveform under consideration. The method is demonstrated on an ECG signal containing an ectopic beat, and an ECG contaminated by muscle artefact noise.