Securing Realistic IoMT Environment Using Moth‐Flame Optimization‐Based Deep Learning Framework

ABSTRACT The Internet of Medical Things (IoMT), a network of medical devices connected by Internet of Things systems, encompasses wearables and remote monitoring tools that collect and exchange data to enhance healthcare delivery. Safeguarding patient information from data breaches on this network is crucial for organizations, as any malicious breach or tampering with the system's functioning can result in significant financial loss. This paper provides a proposed approach focusing on three scenarios: (1) binary attack detection, (2) categorical classification (having 6 categories), and (3) multiclass classification (having 19 classes). The IoMT networks collect data, and a feature selection technique named Moth Flame Optimization is utilized to extract the required features. To do this, a testbed of 40 IoMT devices—25 real devices and 15 simulated devices—was subjected to 18 attacks, taking into consideration the range of protocols used in the healthcare industry. Results reveal that the proposed approach enhances our ability to recognize potential IoMT healthcare devices by presenting a novel method for integrating an artificial neural network with multiple classes.