An Interpretable Adaptive Multiscale Attention Deep Neural Network for Tabular Data

Explanatory AI Predicts the Diet Adopted Based on Nutritional and Lifestyle Habits in the Spanish Population

This study used Explainable Artificial Intelligence (XAI) with SHapley Additive exPlanations (SHAP) to examine dietary and lifestyle habits in the Spanish population and identify key diet predictors. A cross-sectional design was used, employing the validated NutSo-HH scale to gather data on nutrition, lifestyle, and socio-demographic factors. The CatBoost method combined with SHAP was applied. The sample included 22,181 Spanish adults: 17,573 followed the Mediterranean diet, 1425 were vegetarians, 365 were vegans, and 1018 practiced intermittent fasting. Fish consumption was the strongest dietary indicator, with vegans abstaining and some vegetarians consuming it occasionally. Age influenced diet: younger individuals preferred vegan/vegetarian diets, while older adults adhered to the Mediterranean diet. Vegans and vegetarians consumed less junk food, and intermittent fasters were more physically active. The model effectively predicts the Mediterranean diet but struggles with others due to sample imbalance, highlighting the need for larger studies on plant-based and intermittent fasting diets.