Heartbeat classification for detecting arrhythmia using normalized beat morphology features

We propose a method of arrhythmia detection based on beat morphology, which offers a new set of features for heartbeat classification. This can be performed by nearest-neighbor search, which we applied to heartbeats from the MIT-BIH arrhythmia database. Our classifier achieved an overall accuracy of 98.18% on 103,923 heartbeats.