Background Early marriage is widespread around the world and is often recognized as having harmful impacts on young children. In many contexts, it is driven by a combination of structural factors, including economic insecurity and weak legal enforcement. Early marriage is also associated with children’s mortality and health. This study investigates the geographical patterns and socioeconomic determinants of early marriage in Bangladesh and its association with child morbidity. Methods We analyzed data from 8509 ever‐married women aged 15–49 in the 2022 Bangladesh Demographic and Health Survey (BDHS). Binary logistic regression (LR) was used to identify determinants of early marriage (defined as marriage before age 18). Five machine learning (ML) algorithms: LR, support vector machine, decision tree, random forest, and neural network (NN) were applied to assess the impacts of determinants on early marriage. Results Approximately 64% of women in the sample were married before age 18, with the highest prevalence in Rajshahi (75.1%), Rangpur (73.5%), and Khulna (72.1%) divisions. In the adjusted LR model, women’s higher education was the strongest protective factor (AOR: 0.13 and 95% CI: 0.10–0.17). Other significant factors included division of residence, religion, and household wealth. Among ML models, the NN showed the best predictive performance (accuracy: 71.3% and AUC: 70.2%). Conclusions Our study highlights the significant influence of socioeconomic and geographical factors on early marriage among women in Bangladesh. Education emerged as the strongest predictor, underscoring the role of schooling in delaying marriage and empowering women. The application of ML techniques, particularly NNs, yielded valuable insights into classification accuracy, demonstrating their potential to identify at‐risk groups.
Afroja et al. (Thu,) studied this question.