Figure 3: Multiple-dataset optimisation and validation of generic model to fit APs in response to uniform pacing at different frequencies. Experimental APs were obtained by pacing a left atrial intact myocyte at three different pacing intervals (PIs). Model optimisation was repeated three times (runs 1, 2, and 3), each starting at randomized initial parameter values. AP fits obtained for each run were very similar. (a) Three groups of optimised AP fits in response to pacing at intervals (PIs) of 400, 200, and 300 ms. The generic model with five time-dependent currents and one leakage current was simultaneously fitted to the first two datasets (PI = 400 ms and 200 ms) using a single set of model parameters. The optimised model was validated by its ability to reproduce AP responses to pacing at a PI of 300 ms, a dataset not used in the model optimisation. (b) Plots of model generated time-dependent currents for each pacing protocol for each optimisation run. Note the marked AP and membrane current beat-to-beat variations at PI = 200 ms.