IoHT and Cloud-Based Disease Diagnosis Model Using Particle Swarm Optimization with Artificial Neural Networks

Introduction

In general, Internet of Things (IoT) is defined as the process of developing Internet- linked Things over computer systems. IoT states that rather having low power processing devices such as laptops, tablets, and smartphones, it is optimal to have a minimum number of effective gadgets such as wristbands, air conditioners, umbrellas, and refrigerators. Constant human applicable things such as air fresheners and transport has been smartly deployed using processing units, guided by sensors, and produces practical results, which are incorporated in regular devices. Therefore, the linked things are composed of processing as well as communicating abilities by applying tools such as the average lamp or umbrella to link with network communication. The improved objects in IoT have technical reasoning to process the declared operation with no details of a name and feature or character. The domains of technical expertise and electronics have integrated IoHT, an important scientific advancement. IoHT has started to be used in diverse areas.

The term “ubiquitous computing” varies but generally means computing that is available anytime, anywhere and can be processed over a wide range of Internet. The term "Thing” or object represents that the real world is capable of receiving inputs from humans and converts the obtained data to the Internet to process data gathering. For instance, a sewing machine is capable of recording a thread, values of stitches sewn, and the number of stitches the machine can perform. It is more feasible under the application of sensors while recording the function presented by an object with a limited time interval. Actuators could be employed in a sensor node to show the simulation outcome to the real world by linking the objects. These results are enabled by data collected and deal with the Internet.

IoT and cloud computing (CC) have more advantageous features with equal intensity while combining IoT and CC models. The observing method is deployed by integrating two methodologies to track the patient’s data in an effective manner even in remote areas, which is more useful for medical practitioners. IoT schemes often support CC to improvise the function with respect to maximum resource consumption, memory, power, as well as processing ability. Furthermore, CC gains merits from IoT by the improvement of handling the present world and delivers a massive number of novel services in dynamic as well as shared fashion. An IoT-centric CC approach would be expanded for designing fresh methods in the modern world. The concatenation of CC and IoT relying on web fields operates quite well when compared with traditional CC-based domains by means of efficiency.

There are a few novel applications such as the medical, armed forces, and banking sectors that apply the integration of IoT and CC. These combinations would be applicable to provide effective services in medical applications to observe the data from remote sites. IoT-based healthcare domains are employed for collecting required information such as dynamic modifications in health metrics and extend the intensity of medical parameters at the standard time interval. Additionally, IoT tools, as well as medical parameters-based sensor values, are employed effectively to diagnose the disease at the right time prior to attaining a serious condition. As an inclusion, machine learning (ML) techniques are one of the important modules in the decision-making process in the case of large-scale data. The function of applying data analysis for particular regions contributes to data types such as velocity, variety, and volume. The reputed data analysis encloses a neural network (NN), a classification approach, as well as a clustering technique and efficient methodologies.

Data can be produced from diverse sources with specific data types that are more vital in deploying models for handling data features. For IoT, numerous amounts of resources practically develop the required data with no issues of scalability and velocity to find the optimal data method. These factors are assumed to be significant problems of IoT. Here, it is gathered with a large amount of big data that has diverse data such as image, text, and classifying data by applying IoT devices as input data. Such data would be saved in the CC platform with secured healthcare applications. This is employed with a novel ML approach to process the learning function that maps data into two classes: “Normal” and “Disease Affected.”

Diverse works have been carried out by several studies over the last few decades [1]. Verma and Sood [2] established a novel approach to monitor the disease intensity and analyzed under the application of CC and IoT. This is mainly applied for detecting the severity of the disease. The core terms are extended for producing user-relied health values that identify a processing science model. Also, it is employed to observe student health data. In this method, a programmatic health data in student point is produced with the application of a reputed UCI Repository and sensors applied in the medicinal sector to forecast diverse diseases that are affected with severity. It is employed with different classifying techniques to detect diverse diseases. The prediction accuracy is calculated for this model by applying metrics such as F-measure, specificity, and sensitivity. Consequently, it is evident that this method performs better with respect to prediction accuracy when compared to conventional approaches.

Li et al. [3] projected novel energy schemes that operate in end-to-end for CC-based IoT platforms. The energy frameworks are used in examining video stream that is generated by vehicles cameras. It is estimated on the basis of practical testbeds, in particular applications that perform the operations with the application of popular simulators to learn the improvement of IoT devices. Stergiou et al. [4] deployed a review on CC and IoT methods with security problems. Furthermore, it is listed with the contribution of CC and IoT. Consequently, it is illustrated that the duty of CC in IoT functions is the enhancement of applied features. Tao et al. [5] developed a novel multilayer cloud framework to enable efficiency across heterogeneous services that are offered by diverse vendors in the modern home. Also, ontology is included in solving heterogeneity problems that are involved in a layered CC environment. The main aim of ontology is to report the data presentation, knowledge, as w'ell as heterogeneity application that is also used in the security approach to support the security and privacy conservation in interoperations.

Kumar and Gandhi [6] implied a new and reliable three-tier structure to save massive amounts of sensor data. Initially, Tier-1 performs data collection. Secondly, Tier-2 process the large-scale sensor data storage in CC. Finally, a novel detecting technique for Heart Diseases (HD) is developed. As a result, it is carried out with ROC analysis to find signs of HD. Chen et al. [7] applied a new smart in-car camera system that applies a mobile CC method for deep learning (DL). It forecasts the objects from saved videos at the time of driving and selects the specific portions of videos that have to be saved in the cloud for storing local storage space. These models are applicable in attaining optimal prediction value.

Wu et al. [8] focused on developing a novel cloud-centric parallel ML technique for machinery prognostics. It is also employed with a random forest (RF) classifier to detect tool wear in dry milling tasks. Furthermore, a parallel RF method is created under the employment of MapReduce and executed on Amazon Cloud. It is evident that the RF classification model is capable of detecting the exact value. Muhammad et al. [9] performed the monitoring voice pathology of people with the help of CC and IoT. It discusses the possible study of presented voice pathology. Therefore, it is projected with a novel local binary pattern (LBP)-relied detecting system to examine voice pathology inside a monitoring approach. This prediction model attains maximum classification accuracy when compared with alternate traditional methods.

Gelogo et al. [10] defined the fundamentals of IoT with adoptable domains that are accessible in the direction of u-healthcare. This is established with a new technique where it is used in an IoT-centric u-healthcare service. This is helpful in improving the working functions of healthcare care services. Gope and Hwang [11] described a new framework that depends upon IoT medical tools in body area sensors. For instance, a patient could be observed by employing diverse smaller, effective, and lightweight sensors. Furthermore, it is assumed as security requirements for developing the healthcare system. Gubbi et al. [12] explained the vision of structural units as well as upcoming works in IoT technology.

Hossain and Muhammad [13] presented a w'eb observing system named Healthcare Industrial IoT to monitor the patient’s health. This is capable of analyzing patients’ health data to reduce fatality rate. Therefore, it gathers related patient information that is required to examine the application of sensors and medical tools. But embedded in this model to eliminate clinical errors and diverse risk factors are security modules such as watermarking as well as signal enhancements. Zhang et al. [14] presented different methodologies to develop the applications that are accessible to m-healthcare. These website builders are applied for monitoring the patient’s health status under the application of an IoT-centric system. Hence, it has deployed numerous web-relied applications to provide health data of corresponding patients to medical practitioners to provide appropriate treatment. Some other IoT-based healthcare applications are also available in [15-19].

Here, IoT and cloud-based applications are found useful in distinct healthcare applications. To avail prominent e-healthcare services to clients, this chapter presents an IoT and cloud-based disease diagnosis model. A particle swarm optimization (PSO)-based artificial neural network (ANN) called the PSO-ANN model is presented to monitor the diagnosis of the presence of diabetes and its severity level. The application of the PSO algorithm helps to optimize the weights of the ANN model. The data from the benchmark data set and IoT gadgets are used for validation. The validation of the presented PSO-ANN model has been tested using a benchmark diabetes data set. The outcome offered from the experimental analysis clearly pointed out the superior characteristics of the PSO-ANN model over compared methods.

 
Source
< Prev   CONTENTS   Source   Next >