Session S62.1

Development and Validation of a Model of Atrioventricular Conduction in Atrial Fibrillation Based on Junctional Intracardiac Electrograms

A Roka*, B Merkely

Semmelweis University
Budapest, Hungary

Ventricular activation in atrial fibrillation is irregular due to the chaotic atrial electrical activity and filtering effect of the atrioventricular (AV) node. Several models of AV conduction have been proposed, but the ability to predict RR intervals is still poor. Our aim was to develop and validate a new model of AV conduction. Local intracardiac atrial and ventricular electrograms were recorded in 6 dogs near the AV junction. After electrical induction of atrial fibrillation, signals were recorded for 5 minutes or until spontaneous cardioversion. 26406 RR intervals of the first three dogs were used to calculate the RR interval prediction accuracy with each parameter set. 10926 RR intervals of the last three dogs were used to validate the results. Four different timing intervals were used to create the model of AV conduction. An atrial spike in the absolute refractory period was considered as non-conducted and it did not affect the effective refractory period. An atrial spike coming in the effective refractory period was blocked and increased the total refractory period based on its timing. An atrial spike coming in the slow conduction period was conducted to the ventricle with a prolonged AV delay depending on its timing. An atrial event coming in the normal conduction interval was conducted to the ventricle with the normal (minimal) AV conduction delay. An 8-parameter model was created. Using systematic search it was found that the use of absolute or effective refractory period parameters did not increase the accuracy of RR interval prediction. The only parameters that had an effect were the length of slow conduction time interval, the parameters affecting the time-dependent kinetics of AV conduction in this period, and the length of minimal AV conduction time during the normal conduction interval. The RR interval prediction had an average error of 104.7 ms with a standard deviation of 69.8 ms. Validation showed an average error of 101.8 ms and a standard deviation of 58.9 ms. Out model showed that the prediction of RR intervals during atrial fibrillation based on intracardiac recordings near the AV junction is best achieved by using timing-dependent slowing of conduction, instead of using refractory periods.

(Abstract Control Number: 152)