Malicious Medical Portable Document Format Detection using Machine Learning

Attackers are seeking for tools these days that make their jobs easier while keeping the victim unaware of any unwanted activity taking place on his computer. Nowadays online medical reports are shared by healthcare sectors so attackers use medical reports and other medical PDF files for this purpose, inserting malicious code there that later causes harm to the computer and compromise the security of the computer's data. The major goals of this research are to construct the highest quality medical PDF intrusion detection model and to increase the accuracy of the existing research. The objective of the study work is to increase malicious medical PDF detection accuracy and decrease false alarm rates. This study uses the CIC-Evasive-PDFMal2022 dataset to pinpoint malicious or benign PDF file behavior. The machine learning algorithms used in this study model, base learner random forest and adaboost ensemble classifier have accuracy rates of 99.25%, precision rates of 99.34%, and F1 scores of 99.25%. With an accuracy of 99.65%, precision of 99.81%, and F1 score of 99.65%, this is the best quality medical PDF intrusion detection model in comparison to other research in the field. To increase the accuracy, base learner decision tree and adaboost ensemble classifier algorithm were used in the research.