Uncovering Snackification, from Motivations to Behavior: An Analysis of Dutch Millennials Using Mobile Food Diaries

The shift from structured meals to frequent snacking, often referred to as “snackification”, introduces new challenges for mobile health interventions targeting dietary well-being. While mobile Ecological Momentary Assessments (mEMA) enable real-time data collection via food diaries, they often overlook the underlying drivers of snack choices, such as situational motivations (e.g., appetite, socializing) and contextual factors (e.g., location, time). This gap in understanding limits the ability to design personalized and context-aware interventions, which are essential for promoting healthier eating behaviors in real-world settings. To address this gap, we analyzed a longitudinal dataset of 14,312 in-situ self-reports, focused exclusively on snacking and collected over one year from 264 Dutch Millennials. Using association rule mining and machine learning models, we systematically examined two key snack attributes: snack groups and nutritiousness, in relation to multi-dimensional contextual and motivation factors. Our findings reveal associations between specific motivations and snack choices, such as alcoholic beverages aligning with the motivation of sociability ( \(\text{lift}=4.180\) ), and nutritious snacks associating with the motivation of health ( \(\text{lift}=1.926\) ). Notably, dietary motivations emerged as the strongest predictor of snack nutritiousness (nutritious vs. non-nutritious), with machine learning models achieving AUROC scores of \(\approx\) 0.8 via contextual and motivation features. SHAP (SHapley Additive exPlanations) analysis further confirmed that health-related motivations favor nutritious snack consumption, while hedonic ones promote non-nutritious choices. Overall, this study highlights the crucial role of dietary motivations in snack selection and offers practical insights for public health initiatives and mobile food diary design.