Fever Detection Using Convolutional Neural Networks (CNNs)

This research paper is a novel fever detection methodology using the image classification technique with Python-based convolutional neural networks. We have developed a non-invasive and efficient method to identify fever by analysing images of the tongue, based on traditional Chinese medicine. Later on, we built a model which gave 92.2% on the test set with labelled data of images of the tongues. This model obtains better performance from more advanced pre-processing techniques, such as normalization and data augmentation. This study indicates that an integration between ancient diagnostic methods and the latest machine learning algorithms may open new horizons in fever diagnosis during medical practices. Finally, the use of this technology in mobile health applications will promote early treatment, reduce complications, and avoid the need for more complicated interventions.