DeEffNet: A CNN Model for Optimizing the Image Classification of Comorbid Patients

Amidst the increasing global mortality caused by the COVID-19 virus, researchers are committed to finding technological solutions that can support healthcare professionals in their daily responsibilities. Artificial Intelligence (AI) techniques are being employed to provide fast and accurate predictions of disease severity in patients with comorbidities, aiding doctors in their assessment. Currently, X-ray images are used as initial indicators to detect comorbid patients. This research focuses on developing classification models, specifically DenseNet121 and EfficientNetB0. The performance of these models is compared iteratively against a threshold value. The proposed models utilize DenseNet121 and EfficientNetB0 with ReLU activation function and softmax pooling, achieving accuracies of 94% and 77%, respectively. Based on the results, DenseNet121 is considered an efficient classification model.