Temporal reasoning based automatic arrhythmias recognition

We proposed a new temporal reasoning based approach to recognise arrhythmias in real time. Arrhythmias are depicted by chronicle models consisting of a set of events linked by temporal constraints restricting the range of the relative delay between their occurrence time. The temporal reasoner, called a chronicle recognition system, achieves arrhythmia by detecting instances of these chronicle models on the input ECG signal previously transformed into a series of symbolic events. Experimental results demonstrate that the approach is a good complement for the existing methods based on complex QRS classification.