An adapting system for heartbeat classification minimising user input

An adaptive system for the processing of the electrocardiogram (ECG) for the classification of heartbeats into beat classes that seeks to minimize the required input from the user is presented. A first set of beat annotations is produced by the system by processing an incoming recording with a global-classifier. The beat annotations are then ranked by a confidence measure calculated from the posterior probabilities estimates associated with each beat classification. An expert then validates and if necessary corrects a fraction of the least confident beats of the recording. The system then adapts by first training a local-classifier using the newly annotated beats and combines this with the global-classifier to produce an adapted classification system. The adapted system is then used to update beat annotations. Our results show that we can achieve a significant boost in classification performance of the system by using a small number of beats for adaptation