Desktop version

Home arrow Computer Science

  • Increase font
  • Decrease font

<<   CONTENTS   >>

Metaheuristic-Based Kernel Extreme Learning Machine Model for Disease Diagnosis in Industrial Internet of Things Sensor Networks


Nowadays, enhanced utilization of intelligent tools and communication applications in medical observance and the control on actions of medical employees such as doctors, nurses, and hospital managers, patients, and healthcare manufacturers have been witnessed. Based on Gartner and Forbes, it is estimated that the Internet of Things (IoT) has contributed maximum financial cost for the global economy and minimum amount for IoT-based healthcare production [1]. According to these estimates, it can be observed that the Health IIoT is mostly an important player in the Industrial IoTs (IIoTs)-motivated healthcare sector. IIoT has had a remarkable control over several massive and tiny healthcare manufacturing domains. Thus, an enhanced amount of carrying IoT tools, machines, and apps are being utilized to monitor several health-related details such as glucose meter, ECG screens, and blood pressure (BP), etc.

Figure 6.1 shows the industrial IoT healthcare ecosystem.

At present, Health IIoT is in its beginning steps with the consideration for designing and utilization. But IoT-based results are currently showing an extraordinary effect and carving out a developing market. IIoT has the possibility of saving maximum people annually in the United States by eliminating the mortality rate due to the limited medical facilities. It assures patient health and protection by managing vital patient data and synchronizes compared resources (e.g., healthcare staff, services, carrying smart tools for capturing concurrent patient information like critical signs, and patient-compared electronic data) instantaneously with interrelated tools and sensors. It reveals that IoT in the healthcare production makes possible optimal care with costs, diminished direct patient-healthcare staff interface, and ubiquitous allowance for quality care.

Mohammed et al. [2] supposed to design a remote patient observing model utilizing web services and cloud computing (CC). Protected and high-quality healthcare diagnosis is of paramount significance to patients. Consequently, healthcare information protection and patients’ privacy are the significant problems which caused a huge impact on the prospective achievement of Health IIoT. The mainly revolutionary possible function is healthcare observing, as patient healthcare information is gathered from a count of sensors in a network and distributed through healthcare trains to estimate patient care.

Industrial IoT healthcare ecosystem

FIGURE 6.1 Industrial IoT healthcare ecosystem.

A further widespread investigation of IoT in healthcare functions is established in [3]. The IoT-enabled healthcare functions containing IoT-driven ECG observations are explained in the subsequent studies.

Li et al. [4] proposed a health observance as a platform to observe ECG utilizing an adaptive learning examination method for detecting anomalies. Mohammed et al. supposed a remote patient observing model utilizing web services and CC. Particularly, it has planned an Android function to ECG information observing as well as investigation. Information is examined with third-party software when required; but, it is not an opportunity for the cloud server for extracting features. It classifies the signal for assisting the health trains at the time the signal is obtained. In the presented structure, the cloud server removes features and classifies the signal. In order to examine this signal, the decision from the cloud is sent to the healthcare trains for facilitating optimal patient care.

Hassanalieragh et al. [5] explained the options and challenges of health observing and organization utilizing IoT. Several challenges contain slowdown the procedure, managing big data, the occurrence of extra heterogeneous information, and data integrity. During the presented structure, the ECG signal is watermarked on the client-side previous to broadcasting with the Internet for authenticating over some attacks. The data procedure is also shared among the client-side as well as the cloud-side for making the entire method quicker. Jara et al. [6] proposed a remote observing structure utilizing IoT with a presented protocol, known as YOAPY, for creating a safety and scalable fusion of multi-modal sensors for recording critical signs. A cloud-based speech and face identification structure were extended for monitoring a patient’s condition distantly.

Xu et al. [7] offered a ubiquitous data allowing technique in an IoT-based model to emergency medicinal conditions. It has presented a semantic data system for storing information and a resource-based information access technique for gaining management of the data ubiquitously. It is helpful for assisting decision-generating in emergency medicinal conditions. Zhang et al. [8] established a structural design of mobile healthcare networks, which includes privacy-maintaining information gathered and safety broadcasting. The privacy-preserving information gathered is obtained utilizing cryptography through secret as well as private keys. A secure broadcast is obtained utilizing attribute-based encryption where only an allowed user can access the information. This technique is usually useful; but, the major issue is calculation difficulty.

Granados et al. [9] presented web-enabled gateways to IoT-based eHealth through the opportunity to wired or wireless services. For taking benefit of wired gateways with respect to power-efficiency and minimum cost, it has utilized the wired gateways in a tiny room or structure, as progresses are limited. Radio frequency identification (RFID)-based eHealth- care modes were presented. In Amendola et al. [10], a model is projected which captures the patient’s surrounding situations such as temperature and humidity by RFID and broadcasts them for the cloud to a further, brief comprehension of ambient situations. Catarinucci et al. [11] projected an IoT-aware structural design for monitoring and assessing a patient’s condition automatically by combining ultra-high-frequency RFID functions.

Sawand et al. [12] recognized three kinds of threats in an eHealthcare observing model. This is an identity threat, as the identity of the patient is lost or stolen, allowing threat, as an intruder is allowing the model illegally, and disclosure threat, as secret medicinal information are opened using malware or record distributing devices intentionally or unintentionally. It suggested several results for this threat, containing biometric cryptography and an advanced signal procedure method; but, it does not execute the results in their study. A patient with IoT devices and sensors can be used to training a smart healthcare diagnosis model for drug manufacturing and pharmacies (e.g., smart pills, and medicines).

In Chen et al. [13], an emotion-aware or affective mobile computing structures have been projected, and the authors examined the structural design like “emotion-aware mobile cloud” (EMC) for mobile computing. Chen et al. [14] presented another structure, affective interaction through wearable computing and cloud technology (AIWAC). Currently, Flu et al. [15] established the Healthcare IoTs (Health-IoT), trying to bridge among intelligence health observing and emotional care of the patients. So far, it has established no understanding learn on cloud-assisted IIoT-driven ECG observing, as i) an ECG signal is watermarked on the client-side previous to broadcasting with the Internet for the cloud and ii) a cloud server removes features and classifiers the signal for assisting healthcare trains in giving feature patient care.

This paper develops a new Spider Monkey Optimization (SMO)-based Kernel Extreme Learning Machine (KELM) model for disease diagnosis in IIoT environment. The proposed model at first gathers the medical data from heart disease patients utilizing IoT devices. Then, the IoT devices execute data compression by means of Deflate algorithm to diminish the quantity of data being transmitted to a cloud. Concurrently, the cloud server decompresses the data and executes the SMO-KELM model for HD diagnosis. In SMO-KELM model, the parameter tuning of KELM model takes place using SMO algorithm, which depends upon the nature of spider monkeys. Lastly, the alarm module is used to generate an alert in case of the existence of HD. An extensive experimentation will be done to verify the goodness of the presented model.

<<   CONTENTS   >>

Related topics