The Framingham heart study has shown that diabetic patients have an increased risk of cardiovascular disease. Hyperglycemia causes a wide range of multi-scale remodeling mechanisms, but their contribution towards cardiac function deterioration is not well characterized and it may differ between patients. In this study, we performed a one-at-a-time sensitivity analysis on 46 parameters using whole-heart electromechanics models generated from ten UK Biobank datasets (50% females, diabetes at imaging, no comorbidities). To test whether the parameters’ importance was consistent between anatomies, we ranked them according to their sensitivities. The most important parameters for the end-diastolic volume of the left ventricle ( \(EDV_{LV}\) ) were the left heart unloading pressure, the reference pressure of the systemic veins, and the pulmonary resistance, in this order consistently across the cohort. Similarly, the EDV of the right ventricle (RV) strongly depended on the right heart unloading pressure, the reference pressure of the systemic veins, and the systemic resistance. However, the order of the parameters was not consistent between patients, indicating that the RV volume features may be more sensitive to cardiac anatomy. Some of the biomarkers’ sensitivities also differed between sexes (LV end-systolic volume vs systemic resistance sensitivity: male [0.84, 0.99] vs female [0.63, 0.83]). This demonstrates that the male and the female heart may adapt differently to the same remodeling mechanism. Our results show how multi-scale whole-heart models can be used to assess how biomarkers relate to multi-scale cardiac function, and which of these relationships are more sensitive to cardiac anatomy and sex.