Rick Schreurs

109 Left and right ventricular response to CRT RV pressure wire LV pressure wire Baseline A-RV: 125ms Time A-RV: 130ms A-LV: 90ms LV dP/dt RV dP/dt Baseline A-RV: 125ms A-RV: 130ms A-LV: 90ms Time Local EGM (RV) Local EGM (LV) Septum electrodes Model Output RA LA RV LV Baseline A-RV: 125ms A-RV: 130ms A-LV: 90ms LV dP/dt RV dP/dt Activation Early Late Study protocol Measurements Simulations RA LA RV LV Baseline A-RV: 125ms A-RV: 130ms A-LV: 90ms LV dP/dt RV dP/dt Activation Early Late Model output Model input Validation A B C Figure 1. Schematic representation of the methods used in this study. 100 different A-LV/A-RV pacing delay combinations were programmed (A) while pressures and local electrical activation were measured (B). Generic activation maps, derived from local electrograms, were used as onset of mechanical activation in the computer simulations (C). The resulting output of the simulations and measurements was compared for validation purposes. The green square in the heat maps indicates the baseline pacing setting. For both the LV and the RV, response to pacing was defined as the relative change in dP/ dt max compared to the baseline setting. We applied quadratic LOESS fitting to account for measurement variability within each dog [20]. Furthermore, in order to quantify a generic canine response to changes in pacing delay, we created a single representative canine dataset by taking the mean values of the dogs for each setting. We also applied linear 2D interpolation between the measurements in the heat map visualizations of all pacing delay settings. All these calculations were performed in Matlab 2016A (The Mathworks Inc, Natick MA, USA). Computer simulations The CircAdapt model of the entire heart and circulation [21–23] , which can be downloaded from www.circadapt.org , was used to simulate cardiovascular mechanics 6