The healthcare sector requires an instant automated system capable of detecting and assessing the increasing security threats. The solution is based on developing a deep learning model that distinguishes between different types of EXE malware and benign files. The Convolutional Neural Network (CNN) enabled the conversion of EXE files into images by extracting essential visual features. The data splitting process takes place after applying oversampling methods to balance the data before training and adjust image dimensions. A set of dropout controls, combined with the early stopping method within multiple pooling layers, enhances model performance by preventing overfitting. The program achieved 99% accuracy when tested on a dataset consisting of ordinary files and six different malware families. Precision, recall, and F1-score values exceeded 98% for each distinct class. This approach proves to be an effective way to evaluate and identify malware, providing advanced cyber threat protection capabilities in healthcare systems and other domains.