Προγνωστικά μοντέλα επιπέδου χοληστερίνης στο αίμα με βάση δημογραφικά, σωματομετρικά και διατροφικά δεδομένα
Predictive models for the cholestero level in blood on demographic, somatometric and nutrision data
The aim of this paper is to find demographic, somatometric and nutrision factors responsible for the cholesterol level in blood. For this purpose we used data from ATTICA data base. We concentrated mainly on finding the factors responsible for the total cholesterol as well as for the LDL – cholesterol level in blood. More specifically, in Chapter 1 we give the description of the disease, the way the cholesterol level is calculated as well as the methodologies which are used for the statistical analysis. In Chapter 2, the descriptive analysis of the variables is presented. Ιn Chapter 3 the statistical analysis is presented which includes the correlations between cholesterol and all other variables and the tests of statistical hypotheses of the cholesterol level and the other variables( t-test, Chi-squared tests). In Chapter 4 with the use of multiple linear regression, discriminant analysis and logistic regression, predictive models for the cholesterol level in blood are constructed. In Chapter 5 with the use of the Receiver Operating Curve (ROC) we compare the models. In Chapter 6 we summarize our findings .