Many uncertainties remain about the relation between post-infarct scars and ventricular arrhythmia. Most post-infarct patients suffer scar-related arrhythmia several years after the infarct event suggesting that scar remodeling is a process that might require years until the affected tissue becomes arrhythmogenic. In clinical practice, a simple time-based rule is often used to assess risk and stratify patients. In other cases, left ventricular ejection fraction (LVEF) impairment is also taken into account but it is known to be suboptimal. More information is needed to better stratify patients and prescribe appropriate individualized treatments. In this paper we propose to use probabilistic disease progression modeling to obtain an image-based data-driven description of the in-farct maturation process. Our approach includes monotonic constraints in order to impose a regular behaviour on the biomarkers' trajectories. 49 post-MI patients underwent Computed Tomography (CT) and Late Gadolinium Enhanced Cardiac Magnetic Resonance (LGE-CMR) scans. Image-derived biomarkers were computed such as LVEF, LGE-CMR scar volume, fat volume, and size of areas with a different degree of left ven-tricular wall narrowing, from moderate to severe. We show that the model is able to estimate a plausible progression of post-infarct scar maturation. According to our results there is a progressive thinning process observable only with CT imaging; intramural fat appears in a late stage; LGE-CMR scar volume almost does not change and LVEF slightly changes during the scar maturation process.