Computationele tijdreeksanalyse van cardiovasculaire gezondheid en cardiopulmonale fitheidsprofielen

To date, there is an unmet need for accurate computational models that can identify and manage individuals at high risk of developing adverse Cardiovascular (CV) events. The purpose of this doctoral project, is the utilization of machine and deep learning algorithms to create a computational pipeline to construct a profile of patients related to their CV health and fitness and predict their response to potential exercise treatment programmes. The analysis will be performed on time series data, collected from population studies. The specific aims of our project are: (i) to utilize unsupervised learning algorithms on time series Left Attrial/Left Ventricle (LA/LV) longitudinal strain curves to construct left heart deformation profiles; (ii) to describe temporal changes of the LA/LV deformation curves, which will serve as a quantification of how the left heart deformation profile progresses over time; (iii) to investigate whether specific time series clusters of 24-h blood pressure are associated with particular clinical profiles and adverse CV outcome; (iv) to construct of a personalised cardiorespiratory fitness model and an exercise response predictor.