<![CDATA[The impact of the economic crisis after the COVID-19 pandemic and the Russia-Ukraine conflict has affected Indonesia's food stability, including the nutritional intake of school children. To respond to this, this study developed a multivariate logistic regression-based predictive model to identify children's nutritional status (normal, malnourished or obese) based on questionnaire data of parents with children aged 6-15 years. The variables analyzed included eating patterns, physical activity, snacking habits, and consumption of fast food and sweet drinks. The results of the Pearson correlation analysis showed that children with a balanced diet (rich in vegetables, fruits, and protein) tend to have normal nutritional status, while the consumption of fast food andsugary drinks is negatively correlated with nutritional health. Breakfast and physical activity were also shown to have a positive effect. However, the prediction model built has limited accuracy with high specificity for normal cases at 0.92, but low sensitivity at 0.38 in detecting poor nutrition/obesity. This means that the model is not yet reliable enough for critical medical applications, but it remains useful as an initial scoring tool. These findings leadto three practical recommendations: (1) an early detection system for nutrition problems based onsimple scores, (2) personalized nutrition interventions (such as reducing fastfood and increasing physical activity), and (3) refining the Free Nutritious Meal program so that it is well-targeted. Although still in the proof of concept stage (TKT 3), this research provides a scientific foundation for data-driven nutrition policy at the primary healthcare level, especially in the midst of global economic challenges that affect family food security. With this approach, efforts to prevent malnutrition in school children are expected to be more targeted, efficient and adaptive to changing socio-economic dynamics.]]>
