Intelligent Early Diagnosis System against Strep Throat Infection Using Deep Neural Networks

Abstract

The most frequent bacterial pathogen causing acute pharyngitis is Group-A hemolytic Streptococcus (GAS), and sore throat is the second most frequent acute infection. The immunological reaction to group A Streptococcus-induced pharyngitis results in Acute Rheumatic Fever (ARF). A genetically vulnerable host for ARF is a streptococcal infection. ARF, which can affect various organs and cause irreparable valve damage and heart failure, is the antecedent to Rheumatic Heart Disease (RHD). RHD, in many countries is Cardiovascular Disease (CVD) refers to a range of conditions that affect the heart and blood vessels, including coronary artery disease, heart attack, heart failure, and stroke. It is important to note that while this approach has demonstrated promising results, further studies and validation are necessary to establish its clinical feasibility and reliability. Further research can also be done to evaluate the generalization of the model to larger and diverse patient populations. The results showed that using Image Synthesis-based augmentation improved the ROC-AUC scores compared to basic data augmentation. The proposed method could be a valuable tool for healthcare professionals to quickly and accurately diagnose strep throat, leading to timely treatment and improved patient outcomes. The experimental findings indicate that the suggested detection approach for strep throat has a high level of accuracy and effectiveness. The approach has an average sensitivity of 93.1%, average specificity of 96.7%, and an overall accuracy of 96.3%. The ROC-AUC of 0.989 suggests that the approach is effective at distinguishing between positive and negative cases of strep throat. These results indicate that the suggested detection approach is a promising tool for accurately identifying cases of strep throat.