Atrial fibrillation (AF) is the most common cardiac arrhythmia, and its treatment by ablation still presents variable results among patients. In this paper, we present a personalised simulation methodology fitted to clinical data to improve understanding of the atrial electrical substrate and support clinical decision making. We analysed high-density electro-anatomical mapping signals recorded during a triple extra-stimulus protocol able to reveal conduction disturbances on atrial tissue. Computed tomography data was used to build patient-specific 3D atrial models in which electrophysiological simulations were performed and adjusted to reproduce clinical local activation maps (LAT) maps. These included an adjustment of tissue propagation properties and electrical remodelling, to reproduce different levels of disease progression. This methodology reproduced both inter-individual and regional variation in atrial propagation conditions during the triple extra-stimuli. Simulated LAT maps reproduced clinical data with 23.4 ms average deviation, and were better reproduced in patients with lower heterogeneity between LAT maps.