Unveiling socio-demographic determinants of low birth weight using machine learning techniques

Area of Science:

• Maternal and Child Health
• Biostatistics
• Machine Learning in Healthcare

Background:

• Low birth weight (LBW) is a significant global health issue, particularly in low- and middle-income countries like Bangladesh.
• Despite improvements, Bangladesh faces persistent challenges with a 14.5% LBW rate, highlighting maternal and child health disparities.
• Socio-demographic factors critically influence birth weight, necessitating detailed investigation.

Purpose of the Study:

• To identify key determinants of LBW in Bangladesh.
• To develop a machine learning (ML) predictive model for identifying mothers at high risk of delivering LBW infants.
• To inform targeted interventions and policy development for reducing LBW prevalence.

Main Methods:

• Utilized data from the Bangladesh Demographic and Health Survey (BDHS) 2022.
• Applied diverse ML algorithms including Logistic Regression, Naïve Bayes, KNN, Random Forest, SVM, Lasso, Regression Tree, Neural Networks, XGBoost, AdaBoost, and Decision Trees.
• Evaluated model performance using train-test split, 10-fold cross-validation, accuracy, precision, recall, F1-score, R², and MSE.

Main Results:

• 'Age at first birth' and 'Education Level' were identified as the most significant predictors of LBW.
• The AdaBoost algorithm achieved the highest predictive accuracy among all tested ML models.
• The study successfully identified key risk factors and demonstrated the utility of ML in predicting LBW.

Conclusions:

• Age at first birth and education level are critical modifiable factors influencing LBW.
• Machine learning, particularly AdaBoost, offers a powerful tool for predicting LBW risk in vulnerable populations.
• Findings can guide public health policies to mitigate LBW and improve maternal and child outcomes in Bangladesh.