Risk factors analysis and prediction models of obesity in college students based on dietary patterns

Overweight and obesity among college students have become significant public health concerns. This study aims to develop a nomogram model for assessing obesity risk in college students.

A cross-sectional study was conducted among college students in Xuzhou. Demographic, dietary, and lifestyle information was obtained through self-administered questionnaires, while body composition was assessed using the InBody 570 analyzer. Dietary patterns and obesity prevalence were examined through multiple indicators. Principal component analysis (PCA), logistic regression, and a non-invasive risk assessment model based on percentage of body fat (PBF) were applied.

The vegetable meat grain dietary pattern and milk egg dietary pattern were associated with a reduced risk of PBF (P < 0.01), while the snack mode dietary pattern and aquatic meat dietary pattern increased the risk of PBF (P < 0.05). Binary logistic regression identified gender, physical activity, late-night snacking, regular meals, and a healthy diet as key predictors of PBF obesity in college students. The model achieved an area under curve (AUC) of 0.805, with a non-significant Hosmer-Lemeshow (H-L) test (P > 0.05). Decision curve analysis (DCA) showed that the model outperformed extreme curves, indicating its reliability.

This study highlights the high prevalence of overweight and obesity among college students and the importance of using multiple indicators for comprehensive evaluation. The developed PBF-based nomogram model demonstrates potential for obesity screening but requires further validation in diverse populations.