Authors - B Shilpa, Shaik Abdul Nabi, G Sudha Reddy, Puranam Revanth Kumar, Thayyaba Khatoon Mohammad Abstract - The Internet of Medical Things (IoMT) has made it possible for digital devices to collect, infer, and disseminate data related to health through the use of cloud computing. Securing data for use in health care has unique challenges. Various studies have been carried out with the aim of securing healthcare data. The best way to protect sensitive data is to encrypt them so that no one can decipher them. Conventional encryption methods are inapplicable to e-health data due to capacity, redundancy, and data size restrictions, particularly when patient data is transmitted across unsecured channels. Due to the inherent dangers of data loss and confidentiality breaches associated with data, patients may no longer be able to fully protect the privacy of their data contents. These security threats have been recognized by researchers, who have then proposed various methods of data encryption to fix the problem. As a result, the area of computer security is deeply concerned with finding solutions to the security and privacy issues associated with IoMT. This research presents an intrusion detection system (IDS) for IoMT that utilizes the machine learning techniques: Decision Tree (DT), Naive bayes (NB), and K-Nearest Neighbor (KNN). Feature scaling using the minimum-maximum (min-max) normalization method was performed on the CIC IoMT 2024 dataset to prevent information leakage into the test data. The effectiveness of the output was then evaluated, ensuring that the scaling process was correctly implemented as the initial step of this work approach. Five types of assaults are identified in this dataset: DDoS, DoS, RECON, MQTT, and Spoofing. Principal Component Analysis (PCA) was used to reduce dimensionality in the subsequent stage. The suggested methods have a high detection rate of accuracy 98.2%, specificity of 97.6%, recall of 98.0%, and f1- score of 97.8%, which offer a viable option for protecting IoMT devices from attacks.
Open Zoom
