Search academic texts

Information × Registration Number 2124U004691, Article popup.category Препринт Title popup.author Holoborodko Yana popup.publication 01-01-2024 popup.source_user Український католицький університет popup.source https://hdl.handle.net/20.500.14570/4841 popup.publisher Description Dietary patterns and different food groups are important factors in preventing and moderating the progression of a range of chronic diseases. This paper proposes to use data-driven methods in combination with prior knowledge of nutritional epi- demiology 1) to find the most influential factors among food groups that determine population variance in diverse eating habits, 2) to identify dietary patterns, and 3) to find associations of these results with a number of chronic diseases among NHANES 2017 - March 2020 data taking into account, among other things, the respondents’ gender. PCA was used as a dimensionality reduction technique, which showed that the amount of fruits (loading = 0.20 in joint for male and female data), eggs, beans and other non-animal protein (0.16), vegetables (0.26) and grains (0.24) consumed by respondents most explain the variance of dietary patterns in this population. While among men, the presence of alcohol in the diet plays a significant role in differ- entiating as well. Using K-means for clustering, meaningful for interpretation di- etary patterns common to both sexes were named "Whole Foods" (mostly grains, fruits, non-meat protein and dairy products), "Unbalanced" (sandwiches, potatoes and sugars) and "Meat & Alcohol". The associations of dietary patterns and food groups were obtained by using RRR, which proved to be more effective than the usual linear regressions common for this task. The most sensitive to dietary habits in this study were waist-hips ration and high-density lipoprotein, which are among the indicators of obesity and dyslipidemia, respectively. Among the food groups, "Alcohol", "Meat", "Grains", "Sandwiches", and "Snacks&Sweets" had the strongest total associations with chronic disease indicators. Link to the implementation of all the methods can be found here. popup.nrat_date 2025-05-09 Close