AUTOMATIC ATRIAL ARRHYTHMIA DETECTION BASED ON RR INTERVAL ANALYSIS IN CONSCIOUS RATS
Telemetry devices for electrocardiographic (ECG) recording in conscious rats make it possible to obtain long-lasting recordings over extended periods of time. Consequently, the huge amount of data collected with this method raises the need for an automatic arrhythmia detection procedure. Numerous algorithms of arrhythmia detection have been developed for analyzing human ECG data. These algorithms are not well suited for recordings characterized by nonsustained arrhythmias and variable QRS morphology, hence the need for a new algorithm.
In 5 spontaneously hypertensive rats, 24-h ECG recordings were obtained once a month, from 6 to 11 months of age. A common algorithm detected R peaks, and artifacts were discarded either visually or by using an automatic procedure. All atrial and ventricular tachyarrhythmias were visually assessed. Using RR interval time series, automatic identification of tachyarrhythmias was performed by a fuzzy automaton. Parameters of the automaton were optimized using a cross-validation technique and the procedure was evaluated with both visual and automatic artifact detection.
The thirty 24-h recordings yielded a total of 161,615 atrial arrhythmic beats and 5,186 ventricular arrhythmic beats. The automatic detection provided a high sensitivity (0.91) and positive predictive value (0.94) when artifacts were visually discarded, and reasonably good performance (0.89 for sensitivity and 0.91 for positive predictive value) was achieved when using the automatic artifact detection.
This new algorithm is appropriate for telemetric recordings of ECG in rats, and allows a fully automatic, computationally efficient procedure for arrhythmia detection with a good performance.