Statistical modelling and analysis of energy expenditure among severely overweight
Abstract
The purpose of this thesis was to investigate the relationship between the energy expenditure and the FTO genotype (Fat Mass And Obesity-Associated Protein) of 101 severely overweight persons. The statistical methods used involve linear regression with different representations of the energy expenditure as response and the FTO genotype as exposure. Further, other covariates were included in the regression. These covariates were selected using the best subset selection and penalisation through the Lasso approach. The model accuracy was examined through diagnostics plots, and a multi sample-split method was used to determine the $p$-values for the Lasso approach.
When the diet induced energy expenditure was examined, a linear mixed model was used to correctly adjust for the dependencies in the data. A latent class linear mixed model was constructed to look for clustering between the individuals. None of these statistical analyses revealed a relationship between the energy expenditure and the FTO genotype for these 101 severely overweight persons. Sex and weight were the covariates that affected the energy expenditure the most.