MLCNN-CDSE: A Multi-Label Convolutional Neural Network Model for Predicting COVID Drug Side Effects from Images of Stick Structure-Based Chemical Conformers

Predicting COVID drug side effects is essential for ensuring patient safety and optimizing drug treatment strategies. Few drugs have been approved for treating the novel COronaVIrus Disease (COVID). The development of a COVID drug is time-consuming, complicated, and expensive. To address these issues, computational model-based classifiers have been utilized. The existence of side effects in COVID drugs is one of the factors for drug failure. Hence, it is essential to incorporate computational models during the clinical testing phase of the drug. Accurately identifying the adverse effects of COVID drugs is necessary for advancing modern medicine. Therefore, this research presents a Multi-Label Convolutional Neural Network Model for predicting COVID Drug Side Effects (MLCNN-CDSE) using images of stick structure-based chemical conformers. The presented MLCNN-CDSE model is suitable for identifying side effects from stick structure-based drug chemical conformers and outperforms the Inceptionv3, DenseNet201, ResNet50, MobileNetv2, and VGG19 models performance and reporting an accuracy value of 95.97%.