Anomaly Detection-Based Intrusion Detection System Using Deep Neural Networks in Healthcare Internet of Things

Abstract

In recent years, technology has penetrated every domain with every passing second, making things smart. These smart things offer numerous services of convenience to humans and bring in data using various resources [1]. In the State of the Connected Patient report, it is stated that the medical and healthcare Internet of Things (IoT) product review presented that senior care Internet of Things are going to peg up at 119%, as a share of IoT devices. Besides, it is stated that approximately 42% of seniors could start using home monitoring products or wearables within the next three years [2]. Wearables and implants with Internet of Things actuators and sensors help people with their healthcare diagnose a variety of diseases using data from the patient's electrocardiogram (ECG), core temperature, pulse rate, respiratory rate, and oxygen saturation levels as well as information from the patient's workplace computer (such as an iOS or Android smartphone) or implant. [3] . Healthcare relies on the Internet of Things (IoT) for its innovation. With sensors, communication channels, and artificial intelligence (AI), the IoT collects and processes patient data in real-time, generating immediate responses [4], [5]. These systems, up to now, have generally generated much higher levels of trust than they really should [5]. Following old models of security—layering encryption and access controls on top of an interconnected web of smart devices—hasn’t particularly worked out, with the effect of preventing any kind of real, widespread adoption. Traditional, signature-based machine learning (ML) algorithms don’t adapt well to new attack types, either [6]. New attack types are precisely what the IoT in healthcare is inviting, given its somewhat weak security that also reveals the privacy-compromising data that's transmitted to and from a host of healthcare devices. By integrating self-learning procedures and automated capabilities, the suggested deep learning architecture (DLA) centres on smart healthcare anomaly detection. It intelligently controls systems by preprocessing and integrating data from IOT devices in medical settings.