Analyze the Determinants of Malnutrition in Women and Prognosticate Nutritional Status: Insights from the Bangladesh Demographic Health Survey 2022

Malnutrition in women is a significant public health concern and it is a serious issue in Bangladesh. The Bangladesh Demographic Health Survey (BDHS) 2022 was utilized to identify risk variables for malnourished females and fit several machine learning-based approaches to assess their nutritional status. This study included 7972 female individuals of various locations and ages. A chi-square test with a 5% significance level was used to identify possible risk variables for malnutrition in women. Naive Bayes, CART, Logistic Regression, Random Forest, Support Vector Machine, AdaBoost, Extreme Gradient Boosting, and Multilayer Perceptron; these eight machine learning-based classifiers were used to predict malnutrition in women. Summary information revealed that 48.4% of the population analyzed in this study were malnourished women. The chi-square test revealed that fourteen variables were substantially linked with malnutrition in women. Multilayer Perceptron had the highest accuracy of 0.71 for training data but it showed poor performance for the test data set. In terms of efficiency metrics such as accuracy, kappa, and <I>F<sub>1</sub></I> scores, Random Forest outperformed the others. In comparison to the other ML algorithms tested in this study, the Random Forest technique was a significantly effective machine learning-based technique for predicting women's malnutrition in Bangladesh. The proposed approach can help identify high-risk women for malnutrition, reducing the burden on the healthcare system.