یک چارچوب پایش از دور با بازده فوری ترکیبی با الگوریتم NB-WOA
Abstract
1- Introduction
2- Related work
3- Hybrid real-time remote monitoring (HRRM) architecture
4- A case study on patients with blood pressure disorders
5- The Proposed Hybrid Knowledge Discovery Model (HKDM)
6- The proposed NB-WOA algorithm for improving HRRM
7- Results and discussion
8- Conclusions and future work
References
Abstract
The embracing of the Internet of Things (IoT) and Cloud Computing technologies gives excellent opportunities to develop smart healthcare services that have great prediction capabilities. This paper proposes a Hybrid Real-time Remote Monitoring (HRRM) framework, which remote-monitors patients continuously. This smart framework predicts the real health statuses of the patients in real time by using context awareness. The proposed HRRM framework innovates a Patient’s Local Module (PLM) that do a convergence between IoT sensors and clouds. The HRMM transfers some of the computations to the edge of the network in (PLM) such as converting the low-level data to a higher level of abstraction to speed-up the computations in the cloud portion of the HRMM. The convergence of IoT enables the HRMM to use the enormous cloud power in storing, processing, analyzing big data, building classification models for the category of patients’ health status. The local portion of the HRMM uses classification models that have been trained in the cloud to predict the health status of the patient locally in the event of internet interruption or cloud disconnection to save his life in the disconnection periods. Furthermore, this paper proposes a cloud classification technique that is capable of dealing with big imbalanced dataset by minimizing errors especially in the minority class that represents the critical situations. Finally, a hybrid algorithm of Naïve Bayes (NB) and Whale Optimization Algorithm (WOA) has been proposed to select the minimal set of features that achieve the highest accuracy. The (NB-WOA) works as a safe-failure module that decides when to stop the monitoring using HRMM in the case of the failure of influential sensors. Experiments have proved that the HRMM is capable of predicting the health status of the patients suffering from blood pressure disorders accurately. Also, it proved that NB-WOA accelerates the classification process and saves storage space.
Introduction
Machine learning has many contributions in the medical field such as Remote Patient’s Monitoring (RPM) systems that deliver care to the patient suffering from chronic disease especially elderly patients at his home [1]. RPM is defined as using technology to monitor patients remotely (e.g., at his house) to improve patient’s quality of life. It tracks the patient continuously without obstruction to the freedom of his movement to prevent possible complications, and all these services should be provided at reasonable cost [2]. Implantable and wearable biomedical sensors have received much attention over the last two decades because of the need to collect sensor data that contains physiological signals, patient’s activity during vital signs’ measurement, etc. in real time while practicing his daily routine [3]. IoT exploited the progress in ubiquitous sensing which is qualified by Wireless Sensor Network (WSN) technologies to enable actuators and sensors to interact seamlessly with the ambient environment and to share the collected information among different platforms. IoT has made a huge leap by enabling various technologies such as near field communication (NFC), Radio-frequency identification (RFID) and embedded sensor to transform the internet into a fully integrated platform [4,5]. There are many factors that can affect vital signs’ values of the patient such as patient’s activities (current/last), ambient conditions (temperature, humidity, noise, etc.), patient’s habits (sleeping, smoking, alcoholic beverages, food, etc.) and many other factors. Context awareness defines the capability of a system to gather information from the surrounding environment at any time to comprehend it and adapt its behavior accordingly. Context-aware RPM model uses this technique to comprehend the current health situation of the patient and provide a personalized health care service accordingly [6]. For example, context-aware RPM refers to an emergency case when the patient’s heart rate (HR) increases above normal during sleep while refers to a normal case if the increase in HR occurs during exercise.