Predicting High-Risk Pregnancies in India Using NFHS-5 Data: A Machine Learning Study

In India maternal health results are still a major challenge because of lack of proper organisational structure for early diagnosis in rural communities, where high-risk pregnancies often go undetected. This paper aims to provide a machine learning approach to detect risk factors associated with maternal health in its early stages. It uses National Family Health Survey(NFHS 5) data and five supervised learning algorithms including Logistic Regression, Random Forest, Gradient Boosting, Support Vector Machine, and Artificial Neural Network for comparison of performance including accuracy, precision, recall, and F1-score. The model performance demonstrates that the best results were achieved by the Gradient Boosting model with 89% accuracy and 90% recall. The most significant predictors of maternal morbidity included birth spacing, anaemia, maternal age, and education. The results show that machine learning can help improve health programs at an existing level by providing guidance for population-level risk prediction and assisting in policy-based therapy decisions.