Atherosclerosis is a low density cholesterol promoted medical condition in which the walls of the artery thicken due to fatty acid, cholesterol deposition (plaque). Such medical aggravations are known to escalate coronary and cardio-vascular heart diseases (CHD & CVD). This thesis models the time dynamical evolution of atherosclerosis,and in turn coronary heart disease (CHD), as a function of natural ageing and affectation due to life-style parameters like alcohol consumption, cheese consumption,smoking habit, high blood pressure, cereal-fruit-vegtable consumption. Principally based on data modelling (13 European countries, including the UK, have been analysed), followed by a continuum model based prediction, the thesis probabilistically estimates how a change in life style factors could help in controlling CHD/atherosclerosis.
The thesis is structured within three major sections. First, real data from open access databases (WHO & FAO) were analysed using standard statistical tools to establish dependence of CHD rates on the aforementioned lifestyle and ageing parameters.Two major conclusions could be drawn: a) linear dependence of all life style parameters on time, in the post-statin era; b) CHD death rate analysis demarcated the importance of statin usage in medical optimisation of life style factors.
Second, joint variation of (many, if not all) available parameters, including their inter-dependence, was analysed using machine learning based data visualization tools, like Principal Component Analysis (PCA) and NeuroScale (NSC). Two-fold conclusions were drawn: a) low dimensional clustering of high dimensional data established the interdependence of certain parameters; b) a key outcome of this research is the quantification of the moderating influence of the healthy lifestyle factors (fruit/vegetable and cereal consumption) on the negative indicators (systolic blood pressure, smoking, alcohol and cheesy food). This result is expected to lead to a major life saving tool for medical personnel in advising patients on what to eat, how much to eat, and what not to eat.
Combining information from the two sections above, a time varying model was developed that could predict how the population biology data based conclusions could be probabilistically projected to make future predictions of patient behaviour and concerned life expectations related to CHD deaths. This work is presently ongoing.
- data visualisation
- machine learning