AJP - Heart and Circulatory Physiology, Vol 255, Issue 5 1247-H1253, Copyright © 1988 by American Physiological Society
Computer algorithm for modeling transmission over one or two conduction pathways through AV node
L. D. Dorveaux, N. Twidale, N. Robinson and A. M. Tonkin
Department of Cardiology, Flinders Medical Centre, South Australia.
The goal of this study was to develop and test models of conduction through
the atrioventricular node, particularly to encompass possible transmission
over multiple pathways. Output from the atrioventricular node was related
to its input, the variable coupling interval of a single extrastimulus
introduced to scan diastole during programmed atrial stimulation. Trials
were performed three times in 15 patients after pacing for eight beats at
100 beats/min. Hyperbolic and exponential functions were used to model
trial data. For conduction over a single pathway, the standard error of the
estimate was less with the exponential than with the hyperbolic model. The
preferred exponential model had the general form A2H2 = K1 + K2
exp(-A1A2/K3), with K1, K2, K3 being the parameters fitted. The algorithm
describing transmission over two pathways minimized sums of squares of
deviations from two such exponential functions, with potential
"bifurcation" points between the functions being those farthest from the
computed single curve of best fit. The model predicted (R2 greater than
0.90) that 43% of the trials showed single conduction pathways and the
remaining 57% showed dual pathways. A cardiologist who independently
examined the data discarded eight trials because of insufficient data or
scatter and agreed with presence of two pathways and the point of
discontinuity in plots in 81% of trials. The reasons for discordance were
minor and thus the model was reasonably verified.
