Fetal health assessment is a critical component of prenatal care, directly influencing the safety and outcomes of both mother and fetus during pregnancy. Conventional diagnostic approaches, which rely heavily on manual interpretation of cardiotocography (CTG) readings and clinical observations, are often susceptible to subjectivity, latency, and diagnostic inaccuracies, particularly in resource-constrained healthcare environments. This study proposes an AI-driven predictive modeling framework that employs advanced machine learning techniques to classify fetal health conditions into three categories: normal, suspect, and pathological. The system analyzes structured physiological and clinical datasets including fetal heart rate variability, uterine contraction signals, and maternal health indicators. A comparative evaluation of multiple classifiers—including Random Forest, Support Vector Machine, Gradient Boosting, and Decision Tree—was conducted to identify the most effective model. Feature selection and data preprocessing techniques were applied to improve model accuracy and computational efficiency. Experimental results demonstrate that the proposed framework achieves a classification accuracy of 97.8%, with high sensitivity and specificity. This system provides obstetricians and gynecologists with a reliable, automated decision-support tool, reducing diagnostic delays and enhancing prenatal care quality, particularly in rural and under-resourced clinical settings.