Large Scale Cardiovascular Model Personalisation for Mechanistic Analysis of Heart and Brain Interactions

Abstract

Cerebrovascular diseases have been associated with a variety of heart diseases like heart failure or atrial fibrillation, however the mechanistic relationship between these pathologies is largely unknown. Until now, the study of the underlying heart-brain link has been challenging due to the lack of databases containing data from both organs. Current large data collection initiatives such as the UK Biobank provide us with joint cardiac and brain imaging information for thousands of individuals, and represent a unique opportunity to gain insights about the heart and brain pathophysiology from a systems medicine point of view. Research has focused on standard statistical studies finding correlations in a phenomenological way. We propose a mechanistic analysis of the heart and brain interactions through the personalisation of the parameters of a lumped cardiovascular model under constraints provided by brain-volumetric parameters extracted from imaging, i.e: ventricles or white matter hyperintensities volumes, and clinical information such as age or body surface area. We applied this framework in a cohort of more than 3000 subjects and in a pathological subgroup of 53 subjects diagnosed with atrial fibrillation. Our results show that the use of brain feature constraints helps in improving the parameter estimation in order to identify significant differences associated to specific clinical conditions.

Cardiac computational modelling

“…Por ejemplo, se elaboro ´un modelo de la circulacio ´n fetal, personalizado con los datos de ecografı ´a Doppler disponibles, para cuantificar la redistribucio ´n de la sangre en casos de restriccio ´n del crecimiento intrauterino 24 . Los modelos de para ´metros concentrados se han utilizado tambie ´n recientemente para un estudio holı ´stico de las interacciones entre corazo ´n y cerebro, para la personalizacio ´n de modelos cardiovasculares con datos procedentes de miles de casos de la base de datos UK Biobank y para identificar nuevas relaciones entre factores neurolo ´gicos como las lesiones de hiperintensidad de sustancia blanca y anomalı ´as cardiacas como la fibrilacio ´n auricular (FA) 25 . En este sentido, tambie ´n es posible crear un modelo simple basado en el acoplamiento de modelos 1 D o 0 D con modelos 3 D para estudiar de manera eficiente problemas de hemodina ´mica 26 .…”

Modelización computacional cardiaca

“…24 Lumped models were also recently employed to study heart-brain interactions in a holistic way, personalizing cardiovascular models with data from thousands of cases from the UK Biobank dataset, finding new relations between neurological factors such as white-matter hyperintensity lesions and cardiac abnormalities such as atrial fibrillation (AF). 25 In that line, it is also possible to create a reduced order model based on coupling 1D or 0D models with 3D models to study efficiently hemodynamics problems. 26 1D models allow us to investigate physical mechanisms underlying changes in pressure and flow pulse waveforms that are produced by a cardiovascular disease 27 .…”

Joint data imputation and mechanistic modelling for simulating heart-brain interactions in incomplete datasets

“…The parameters of the model were selected based on the available clinical data. To obtain the target values for the emulator, the data-assimilation procedure was carried out according to the approach presented in [2,12].…”