<![CDATA[Early detection of dengue fever remains a clinical challenge without prior laboratory examination, due to its overlapping symptoms with other febrile conditions, most notably high body temperature. By observing these symptom similarities, this study proposes a comprehensive comparison of machine learning algorithms using a classification-based early detection approach with Random Forest, Gradient Boosting, Support Vector Machine, and Decision Tree. The dataset was obtained from a regional hospital with ethical clearance, consisting of 212 records collected in January 2025, with 11 selected clinical features including body temperature, fever duration, pain, nausea, vomiting, cough, flu, rash, headache, nosebleed, and gum bleed. The models were evaluated using accuracy, precision, recall, and F1-score, both before and after hyperparameter tuning. GridSearchCV was applied to identify the optimal hyperparameter combination for each model. In this study, recall is prioritized as the primary evaluation metric to ensure the model performs well in minimizing missed dengue cases. The results demonstrated that SVM achieved the best performance across all metrics except precision, both before and after tuning. These findings suggest that SVM is the most suitable model for clinical early detection of dengue fever using symptom-based data.]]>
Copyrights © 2026
