Abstract: Amid the escalating global mortality stemming from the COVID-19 virus, researchers are dedicated to exploring technological innovations to bolster the efforts of healthcare professionals. Artificial Intelligence (AI) techniques are being harnessed to swiftly and accurately predict disease severity in pa- tients with comorbidities, thereby assisting healthcare providers in their evaluations. Presently, initial detection of comorbid patients relies on X-ray images. This study centers on the development of classification models, specifically DenseNet121 and NANSNetLarge. The performance of these models is sys- tematically compared against a predetermined threshold value. The proposed models leverage DenseNet121 and NANSNetLarge with ReLU activation function and softmax pooling, resulting in accuracies of 95% and 81%, respectively. Based on the findings, DenseNet121 emerges as an effective classification model.
Index Terms: Comorbid, COVID-19, DeanseNet121, NANSNetLarge ReLU, Softmax pooling.
|
DOI:
10.17148/IJARCCE.2024.13547
