Type 2 diabetes causes adverse cardiac remodelling at subcellular, tissue, and whole-organ level. This can lead to severe cardiac complications if left uncontrolled, notably cardiac arrhythmias and heart failure. A timely identification of cardiac abnormalities and a rigorous understanding of underlying mechanistic causes are therefore crucial to prevent further cardiac deterioration in patients with type 2 diabetes. Our goal is to identify and mechanistically explain subclinical abnormalities in the diabetic heart, by combining Big Data analysis at population-level followed by patient-specific multi-scale modelling and simulation. Rich in quantity and diversity of data, the UK Biobank is an ideal resource to quantify those changes in such a high-risk yet understudied population before the development of overt cardiac disease. While cohort studies do not offer mechanistic insight, modelling and simulation approaches provide a reliable virtual testbed to investigate possible pathophysiological mechanisms underlying those changes. In this work, we present a computational framework to characterise cardiac remodelling in type 2 diabetes in the absence of cardiovascular disease, using human-based multi-scale modelling and simulation to explain possible mechanisms underlying the differences observed in clinical biomarkers at population level.