Supervised machine learning-based prediction of modern contraceptive use among sexually active women in Nepal

Modern contraceptives play a key role in reducing maternal and infant mortality, preventing unsafe abortions, and advancing women’s social and economic participation. Despite their importance, contraceptive use remains uneven across subpopulations in Nepal. This study aimed to identify key predictors of modern contraceptive use among sexually active women using supervised machine learning and data from the 2022 Nepal Demographic and Health Survey. A total weighted sample of 7,090 women residing with their partners was analyzed. Four supervised machine learning models, Random Forest, Extreme Gradient Boosting, Support Vector Machine, and Logistic Regression, were trained on a balanced dataset. Hyperparameters were optimized through Bayesian optimization. Model performance was assessed on an independent test set. The area under the receiver operating characteristic curve was model evaluation metric. Shapley Additive Explanations quantified predictor importance. The prevalence of modern contraceptive use was 53.70%. Random Forest model achieved the best predictive performance, with highest accuracy and an area under the curve. The most important predictors were fertility preference (0.0737), children ever born (0.0291), secondary or higher education (0.0214), internet use (0.0211), having at least one son (0.0203), contraceptive decision-making (0.0195), Janajati ethnicity (0.0181), age 35–49 years (0.0176), current employment (0.0169), and pregnancy loss history (0.0151). Women who did not desire another child, had children, used the internet, were employed, belonged to the Janajati ethnic group, aged 35–49 years, or had experienced pregnancy loss were more likely to use modern contraceptives. Women who desired another child, lacked formal education, or made contraceptive decisions independently showed a lower likelihood of use. Thus, modern contraceptive use is jointly shaped by demographic, reproductive, and socioeconomic factors. Interventions should prioritize younger women and adolescents, expand digital reproductive health communication, encourage couple engagement in contraceptive decision-making, and implement culturally tailored programs to address disparities across sub-populations.