Table 1.
The first, second and third columns show the parameter name, range or value, and its meaning. The last column provides the original paper the symbol refers to. The blue rows indicate parameters that were estimated from clinical data, while the gray rows represent parameters the model outputs were insensitive to. HM in the second column indicates the parameters that were constrained by a history matching procedure on a sub-model (see S2, S3 and S4 Files). Abbreviations/symbols: CV = conduction velocity, FEC = fast endocardial conduction, Na+ = sodium, K+ = potassium, Ca2+ = calcium, Cl- = chloride, SR = sarcoplasmic reticulum, DS = diadic space, HM = history matching.
Table 2.
Simulator parameters (continued).
Abbreviations/symbols: Ca2+ = calcium, CaTRPN = calcium/troponin complex, LV = left ventricle, RV = right ventricle, fraction of unbound (U), and weakly (W) or strongly (S) bound binding sites.
Table 3.
Simulator parameters (continued).
Fig 1.
Left: patient-specific four-chamber heart mesh. Right: refines atria, atrial and ventricular transmural myofiber orientation from endocardium (red) to epicardium (blue).
Fig 2.
Electrophysiology simulations.
Left: atria with the region representing the Bachman bundle (red) and the atrial activation site (light-blue circle). Right: ventricles with the fast endocardial conduction layer (blue) and the ventricular activation site (orange circle).
Fig 3.
Mechanics boundary conditions.
A Penalty map scaling the normal spring stiffness for the effect of the pericardium. B Afterload and preload boundary conditions represented with CircAdapt. Symbols and abbreviations: p = pressure, R = resistance, q = flow across a valve, LV = left ventricle, RV = right ventricle =, LA = left atrium, RA = right atrium, Ao = aorta, Pa = pulmonary artery, Ve = veins, PVe = pulmonary veins, sys = systemic, pulm = pulmonary, MV = mitral valve, TV = tricuspid valve, AV = aortic valve, PV = pulmonary valve.
Fig 4.
Orange, blue and purple boxes indicate Latin hypercube sampling, simulator evaluation and GPE training, respectively. Blue and purple X indicate points where the simulator or the GPEs are evaluated.
Fig 5.
Summary of sub-models analysis.
The diagram shows a summary of the analysis performed on the sub-models before performing the GSA on the four-chamber electromechanics model.
Fig 6.
Simulated pressure-volume transients.
Top: LV pressure-volume transients and pressure traces simulated with the fully-coupled simulator (black), with the orange and the blue circles representing the resulting LV peak pressure and transient duration. The horizontal and the vertical lines show the clinically measured average LV peak pressure and transient duration, respectively, to show that the simulator achieves physiological values. Bottom: RV, LA and RA simulated pressure-volume transients.
Table 4.
Average R2 score and ISE over the five cross-validation splits, reported for each output.
Fig 7.
Total effects of input simulator parameters (y-axis) on model outputs (x-axis) normalised between 0 and 1 for each output.
Fig 8.
The effect of simulator parameters on ventricular diastole and atrial systole.
Barplots of the total order effects of the input parameters, ranked from most to least important, for ventricular EDV, EDP and atrial ESV and peak pressure. The red and blue bars indicate parameters that explain >90% of the output variance, the gray bars indicate parameters that explain up to 95% of variance, while the rest of parameters are not displayed. Red and blue indicate positive and negative interactions, respectively. When this resulted in more than 10 parameters, the number of parameters was limited to 10 or to the number of parameters explaining 90% of output variance.
Fig 9.
The effect of simulator parameters on ventricular systole and atrial diastole.
Barplots of the total order effects of the input parameters, ranked from most to least important, for ventricular peak pressure, ESV and atrial maximum volume during the v-wave. The red and blue bars indicate parameters that explain >90% of the output variance, the gray bars indicate parameters that explain up to 95% of variance, while the rest of parameters are not displayed. Red and blue indicate positive and negative interactions, respectively. When this resulted in more than 10 parameters, the number of parameters was limited to 10 or to the number of parameters explaining 90% of output variance.
Fig 10.
The effect of simulator parameters on ventricular pressure rise and decay rate.
Barplots of the total order effects of the input parameters, ranked from most to least important, for ventricular maximum and minimum derivatives. The red and blue bars indicate parameters that explain >90% of the output variance, the gray bars indicate parameters that explain up to 95% of variance, while the rest of parameters are not displayed. Red and blue indicate positive and negative interactions, respectively. When this resulted in more than 10 parameters, the number of parameters was limited to 10 or to the number of parameters explaining 90% of output variance.