An Explainable Artificial Intelligence–Driven Hybrid Clinical Prediction Model for Accurate Gestational Age and Expected Delivery Date Estimation in Resource-Limited Settings

Accurate estimation of gestational age (GA) and expected date of delivery (EDD) is vital for quality antenatal care, obstetric decision-making, and improved maternal–fetal outcomes. In low-resource settings, limited access to reliable ultrasound and early prenatal assessment often forces clinicians to depend on less accurate methods such as last menstrual period (LMP) and fundal height. This study developed and validated an explainable artificial intelligence (AI)-driven hybrid clinical prediction model to improve GA and EDD estimation by integrating maternal history, obstetric variables, and available ultrasound biometric data. The study recruited 1,200 pregnant women from selected public and mission hospitals. Data collected included demographic characteristics, LMP, fundal height, parity, obstetric history, and ultrasound parameters, including biparietal diameter, femur length, abdominal circumference, and head circumference. A hybrid predictive framework combining Bayesian data fusion and machine learning algorithms was used to estimate GA and predict EDD. Performance was assessed using mean absolute error, root mean square error, sensitivity analysis, Bland–Altman agreement analysis, and k-fold cross-validation. The model achieved a mean absolute error of 5.9 ± 3.8 days for GA estimation and an EDD prediction error of 5.2 ± 4.1 days, outperforming LMP- and fundal height-based approaches. It also showed strong sensitivity for preterm, term, and post-term classification, with stable performance among late antenatal attendees and women with incomplete ultrasound records. Explainability analysis improved interpretability and clinical trust. Overall, the findings show that AI-driven hybrid models can provide practical, scalable, and cost-effective decision support for antenatal care in underserved communities.