Dengue constitutes a vector-borne viral infection common in developing countries that presents diagnostic challenges for healthcare providers. The present work established interpretable AI systems to forecast dengue fever from laboratory data for clinical confidence. A total of 1,523 patient records were used in a comprehensive analysis, where six algorithms were evaluated: XGBoost, Random Forest, Decision Tree, Multilayer Perceptron, Support Vector Machine and K-Nearest Neighbors. Advanced preprocessing incorporated hematological ratios, SMOTE for class balancing, and standardized scaling. Model interpretability was enhanced using SHAP and LIME techniques. XGBoost demonstrated superior performance, achieving 86.71% accuracy, 0.87 precision, 0.87 recall, 0.87 F1 score, and approximately 0.90 AUC. The most influential predictive biomarkers according to SHAP analysis were monocyte percentage, neutrophil percentage, and total platelet count, consistent with established clinical knowledge of dengue pathophysiology. Our study showed that routine hematological parameters enable machine learning model development for accurate dengue disease predictions. Explainable AI integration will help to develop actionable insights for real-world implementation, providing an inexpensive diagnostic solution of clinical utility, especially suited for resource-poor, endemic countries.