Search

Background and Objectives: Malnutrition among women is a major public health problem that has been linked to stunted growth, diabetes, and has adverse consequences for children, including low birth weight, less resistance to infections, and a higher risk of death. This current study presents an exhaustive comprehensive study of machine learning (ML) system which has two major objectives: (i) identification of the potential risk factors of malnourished women; and (ii) propose a better ML-based model for predicting malnourished women. Methods: About 15,464 respondents were taken from the Bangladesh Demographic and Health Survey. The potential risk factors for malnutrition were extracted using multinomial logistic regression (MLR). Five ML-based algorithms such as Naïve Bayes, support vector machine, decision tree, artificial neural network, and random forest (RF) were implemented for predicting malnourished women and evaluating their performances using accuracy and area under the curve (AUC). Results: MLR illustrates that age, region, wealth index, respondent's education, currently breastfeeding and pregnant women, marital status, toilet facility, and cooking fuel are the potential risk factors for underweight, while age, region, residence, wealth index, respondent's and husband's education, currently breastfeeding and working, children's ever-born, drinking water source, and cooking fuel for overweight/obese. Our findings show that RF-based classifier provides 81.4% accuracy and 0.837 AUC for underweight and 82.4% accuracy and 0.853 AUC for overweight/obese. Conclusion: The combination of MLR-RF-based system could accurately classify malnourished women with higher accuracy. The proposed system will be helpful in predicting which women are at high risk of malnutrition and reducing the burden of the health system.

The objective of this study was to investigate how higher secondary students perceive the online and to identify the factors that affect their perceptions during the Covid 19 epidemic in Bangladesh. The research is quantitative in nature. 1078 students' data were gathered using an online survey questionnaire through multistage random sampling. The results demonstrate that three factors—online learning's challenges and difficulties, the effectiveness of online learning, and students' motivation for online learning—have a substantial impact on how students view online learning. Despite various difficulties using online platforms, students have a favorable attitude toward online learning.