Abstract
Maternal mortality remains a critical public health challenge in low- and middle-income countries, where delayed diagnosis and inadequate risk stratification contribute significantly to preventable adverse outcomes. This study proposes an artificial intelligence (AI)-based predictive modeling framework for early detection of high-risk pregnancies in resource-constrained healthcare systems of Pakistan. A quantitative, cross-sectional research design was adopted using secondary clinical data comprising demographic, clinical, and obstetric variables. Multiple machine learning algorithms, including Logistic Regression, Support Vector Machine (SVM), Random Forest, and Gradient Boosting, were trained and evaluated using a 70:30 train-test split. Model performance was assessed using accuracy, precision, recall, F1-score, and ROC-AUC. Results indicated that ensemble learning methods outperformed traditional statistical approaches, with the Gradient Boosting model achieving the highest predictive accuracy (93.6%) and ROC-AUC of 0.96. Key predictors of high-risk pregnancy included blood pressure, body mass index, hemoglobin level, and gestational history. The findings confirm that AI-based predictive systems significantly enhance early risk identification and support clinical decision-making under resource limitations. The study concludes that integrating machine learning–based decision-support systems into maternal healthcare can improve early diagnosis, optimize referral processes, and potentially reduce maternal morbidity and mortality in Pakistan. The proposed framework offers a scalable and context-specific solution for strengthening digital health infrastructure in developing healthcare systems