Insights From the National Survey Leveraging Machine Learning and SHAP to Identify and Predict Associated Factors of Malnutrition Among Bangladeshi Women
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
This study investigates the dual burden of malnutrition—underweight and overweight/obesity among Bangladeshi women through the application of traditional statistical techniques and machine learning (ML) models. We applied Multinomial Logistic Regression (MLR) alongside ML classifiers, including Random Forest (RF), Extreme Gradient Boosting (XGB), Gradient Boosting Machine (GBM), Support Vector Machine (SVM), and Multilayer Perceptron (MLP). Model performance was evaluated using classification metrics: accuracy, precision, recall, F1‐score, and area under the ROC curve (area under the curve, AUC). SHapley Additive exPlanations (SHAP) analysis was used to enhance model interpretability by identifying the most influential predictors. The findings reveal distinct socio‐demographic and regional patterns associated with malnutrition. MLR reveals that underweight status is more prevalent among younger women, those with low education, lower household wealth, rural residence, and larger family size. Conversely, overweight/obesity is more common among older, wealthier, urban, and more educated women, especially those who are not breastfeeding or engaged in formal employment. Among the ML models, ensemble‐based approaches such as RF and XGB demonstrated superior predictive performance across all evaluation metrics. SHAP analysis confirmed that wealth index, age, education level, media exposure, and reproductive factors were the most influential variables in determining nutritional status. Furthermore, SHAP visualizations uncovered nonlinear and context‐specific associations that complemented traditional regression findings. These results highlight the urgent need for tailored public health interventions that simultaneously address undernutrition in disadvantaged populations and overnutrition in affluent urban groups. The integration of interpretable ML offers a powerful framework for developing data‐driven, equitable nutrition policies in Bangladesh.
Fields of Study & Subjects
Artificial Intelligence in Healthcare
Obesity
Physical Activity
Diet
Public Health and Nutrition
Publication details
- Journal / source
- Food Science & Nutrition
- Volume
- 14
- Issue
- 4
- Pages
- e71687
- Publication date
- 30 March 2026
- Language
- EN