<![CDATA[This study addresses the challenge of diagnosing Lyme disease through automated classification of Erythema Migrans (EM) rashes, a primary symptom. Employing a Voting Classifier within a k-fold (k=5) cross-validation framework, we developed and validated a model based on a curated dataset of EM rash images and similar dermatological conditions. Image pre-processing involved segmentation and feature extraction using Hu Moments, preparing the data for effective machine learning application. The classifier demonstrated an average accuracy of 81.37%, with variations in precision, recall, and F1-scores across folds, indicative of the model’s robustness and areas for improvement. The results suggest that while the Voting Classifier is a promising tool for Lyme disease diagnosis, further enhancements are required to optimize its diagnostic performance fully. Significant research contributions include the development of a publicly accessible EM rash dataset and the application of ensemble learning techniques to medical image classification, offering a foundation for future advancements in automated disease diagnosis. Recommendations for ongoing research include expanding the dataset diversity and integrating multi-modal clinical data to enhance model accuracy and applicability.]]>
Copyrights © 2024
