[en] Free-living individuals have multifaceted diets and consume foods in numerous combinations. The effect of the overall diet beyond that of single foods can be studied with dietary pattern analysis. Furthermore, the dietary pattern approach reduces data-dimensionality and alleviates problems of model over-fitting and residual confounding that occur with the statistical analysis of many food items.
Most recent dietary pattern analyses have used factor and cluster analysis. We describe a finite mixture modelling (FMM) approach for dietary pattern analysis and show its advantages over previous ones.
First, FMM allows estimating pattern prevalence directly from the model parameters as opposed to the subjective joint classification of the factors. Moreover, in contrast to 'hard' assignment of clustering methods, FMM also produces posterior cluster membership probabilities for each subject providing measures of uncertainty of the associated classification.
Second, it allows problems in determining the number of clusters and choosing an appropriate clustering method to be recast as statistical model choice problems.
Third, it allows for covariates adjustment simultaneously with the fitting process and the size of pattern to depend on a set of concomitant variables. Additionally, FMM is invariant to linear transformation, for example standardization.
We discuss these advantages and illustrate the approach with an analysis of the NESCAV (Nutrition, environment and cardiovascular health) dataset (Alkerwi et al, 2010) and show how identified dietary patterns and their associated uncertainty can be used to predict disease.
Research Center/Unit :
Competence center in methodology and statistics
Disciplines :
Public health, health care sciences & services
Author, co-author :
Sauvageot, Nicolas ; Université de Liège - ULiège > Form. doct. sc. santé publique
Albert, Adelin ; Université de Liège > Département des sciences de la santé publique > Département des sciences de la santé publique
Alkerwi, Ala'a
Guillaume, Michèle ; Université de Liège > Département des sciences de la santé publique > Epidémiologie nutritionnelle
Language :
English
Title :
Use of finite mixture models for dietary patterns analysis
Publication date :
25 August 2014
Event name :
35th annual conference of the international society of clinical biostatistics.
Event organizer :
ISCB
Event place :
Vienna, Austria
Event date :
from 24-08-2014 to 28-08-2014
Audience :
International
References of the abstract :
international society of clinical biostatistics. Number 57. June 2014