Submitted Abstract
Background: Heart failure is the number one reason for death in the EU. It is a progressive disorder with multiple aetiologies, (e.g. volume-pressure overload, ischemic heart disease), subforms (HFpEF, HFrEF) and severities. Patients can be asymptomatic and go undetected and undertreated for years increasing the risk of adverse outcome. Mechanistic physiology based models have matured to clinical application enabling personalized assessment of the underlying pathophysiological processes. While the availability of multidimensional bio-medical data on individual patients (imaging, sensors, omics) has increased dramatically in recent years allowing a progressively fine-granular classification of patients by data driven models, mechanistic models rely on very specific high quality data, that is not always available the clinic (problem of missing data).Main objective: The main objective is to enable patient specific modeling for improved diagnosis and clinical decision making in cardiovascular medicineMethods: We propose a novel concept (the HeartMed platform) that in a step-wise approach combines data-driven and mechanistic modeling# Deep-phenomapping will be done for selected animal models and human patients using data from imaging, omics, sensors# Heart failure subclasses assessing conformities and differences between the pre- and clinical phenotypes will be defined# Based on the phenomapping classification, we will combine pre-/clinical information to impute missing data and enable the application of physiology based mechanistic models of myocardial metabolism and circulation for patient specific modelling# Patient specific models will be used for the assessment of patient individualized cardiac functionality to improve diagnosis and help clinical decision-making# In a Clinical proof-of-concept study of patients with heart failure, we will use HeartMed to compare model driven patient classification, diagnostics and treatment with current clinical procedures.