Large‐scale analysis of heart‐brain interactions through personalisation of a mechanistic cardiovascular model

Background Heart and brain are linked by means of pathophysiological and physiological mechanisms sharing several risk factors (Doehner, W. et al. 2018). Clinical evidence supports a common underlying cardiovascular pathophysiology relating cardiac function and brain damage (Moroni, F. et al. 2018). Nevertheless, heart and brain have usually been studied independently and until now the lack of databases combining data from both has prevented to properly study their relationship. Computational cardiovascular models can complement statistics imposing prior knowledge using a mechanistic approach, and can help us to estimate non‐observable parameters key for the cardiac function such as cardiac contractility or aortic compliance, which cannot be measured in‐vivo. Method extending the approach presented in (Mollero, R. et al. 2018) we propose to constrain the personalisation of a cardiovascular model to ccount for the relationships between cardio‐vascular parameters and brain‐volumetric features extracted from imaging data. We applied our framework in a cohort of more than 3400 subjects and in a pathological subgroup of 59 subjects diagnosed with atrial fibrillation (AF). Results We observed an association between non‐observable parameters such as aortic stiffness or peripheral resistance with brain‐volumetric features such as white matter hyperintensities (WMH) and brain ventricles volume. Moreover, we observed differences in the estimated parameters between AF subjects and control groups linking AF with WMH and cardiac remodeling (Seko, Y. et al. 2018). Conclusion Our methodology allows to study the influence of the brain features in the estimation of the cardiac function parameters, and to identify significant differences associated to specific clinical conditions, such as between atrial fibrillation and white matter damage.