Early Prediction of Adhd Using Machine Learning Techniques In Educational Contexts With Limited Resources: Case of San Pedro Sula, Honduras

Early detection of Attention Deficit Hyperactivity Disorder (ADHD) is critical for effective intervention, yet current diagnostic methods are often time-consuming and resource-intensive. This study explores the application of machine learning techniques to predict ADHD among 81 students from San Pedro Sula, Honduras, using a dataset with 38 variables that encompass demographic, clinical, and behavioral information. Multiple classifiers were evaluated, including Logistic Regression, Modified Logistic Regression, Random Forest, Support Vector Machine, Naive Bayes, F1 Tuned Naive Bayes, and an Ensemble Voting Classifier. In addition, a Naive Bayes model with selected features—identified through a pre-optimized Random Forest—was developed. Among the models, the NB with Selected Features model achieved the best balanced performance, with an accuracy of 0.74, a sensitivity of 0.87, a specificity of 0.46, and an AUC of 0.77. These results underscore the importance of targeted feature selection in small, high-dimensional datasets, and demonstrate that a simple probabilistic classifier can be effectively employed for early ADHD detection in resource-constrained environments. Future work will focus on refining feature engineering and threshold calibration to further enhance model performance and generalizability.