Traditional and Non-Traditional Clustering Techniques for Identifying Chrononutrition Patterns in University Students

Area of Science:

• Chrononutrition and cardiometabolic health.
• Application of computational methods in nutritional science.

Background:

• Chrononutrition, the study of food intake timing relative to circadian rhythms, is crucial for cardiometabolic health.
• Limited research has explored integrated meal timing patterns and compared clustering approaches for their identification.

Purpose of the Study:

• To compare four clustering techniques (K-means, Hierarchical, Gaussian Mixture Models, Spectral) for identifying habitual meal timing patterns.
• To characterize identified patterns using sociodemographic, anthropometric, food intake quality, and chronotype data.

Main Methods:

• Cross-sectional study of 388 Mexican university students.
• Application and comparison of traditional (K-means, Hierarchical) and non-traditional (GMM, Spectral) clustering methods.
• Assessment of clustering method concordance using Adjusted Rand Index (ARI).

Main Results:

• Five distinct meal timing patterns were identified: Early, Early-Intermediate, Late-Intermediate, Late, and Late with early breakfast.
• Chronotype correlated with meal timing patterns, with morning types in earlier clusters.
• Food intake quality varied across patterns, with early eaters exhibiting healthier intake than late eaters.
• Moderate concordance was observed across clustering methods (mean ARI = 0.376).

Conclusions:

• Both traditional and non-traditional clustering techniques identified similar core structures in meal timing patterns.
• Comparing multiple clustering methods is essential for robust chrononutrition research.
• Future research should explore these patterns in diverse populations and their link to cardiometabolic outcomes.