Abstract
This study developed and evaluated a machine learning-based predictive framework for the early detection of malnutrition, growth disorders, and developmental delays among Pakistani children. The objective was to enhance early diagnostic accuracy by integrating multidimensional predictors, including anthropometric indicators, maternal health status, dietary diversity, and socioeconomic conditions. A quantitative and computational research design was employed using secondary pediatric health datasets, which were processed and analyzed through multiple machine learning algorithms, including Logistic Regression, Support Vector Machine, Random Forest, Gradient Boosting, and Artificial Neural Networks.The models were trained and validated using a stratified dataset split and ten-fold cross-validation to ensure robustness and generalizability. Performance was assessed using accuracy, precision, recall, F1-score, and ROC-AUC metrics. The results indicated that ensemble learning and deep learning models significantly outperformed traditional statistical approaches, with Artificial Neural Networks achieving the highest predictive performance. The findings further revealed that socioeconomic status, BMI-for-age, maternal health, and dietary diversity were the most influential predictors of child health outcomes.The study concluded that machine learning-based systems provide an effective and scalable solution for early detection of pediatric malnutrition and developmental delays in resource-limited settings such as Pakistan. The proposed framework has strong potential to support early intervention strategies, improve pediatric healthcare outcomes, and strengthen national child health surveillance systems.