Automated Fever Detection by Ensemble Learning Technique: An Investigative Study

Fever is a common symptom for many infectious and inflammatory conditions. Manual microscopy diagnosis of fever is time-consuming and error-prone. This paper investigates automated ensemble learning approaches for classifying fever using cell images. Methods like discrete cosine transform (DCT), principal component analysis (PCA), gray level co-occurrence matrix (GLCM), and local binary patterns (LBP) are evaluated for feature extraction. The features are used to train ensemble classifiers including AdaBoost, random forest, and gradient boosting machine (GBM). Experiments indicate GBM provides the highest accuracy of 95.2% with GLCM features. Deep convolutional neural network (CNN) based features further improve accuracy to 96.8%, albeit with higher computation time. The results demonstrate the feasibility of automated fever detection from cell images using ensemble learning.