Research On Remote Monitoring System For The Elderly And Its Health Prediction Model

China’s population aging process is deepening,the problem of elderly guardianship is highlighted,and it is important to study the remote guardianship system for the elderly.Io T has been one of the catalysts to revolutionize the traditional means of monitoring,providing new ideas for remote monitoring of the elderly,but the flexible deployment and accurate analysis of data is still a challenge for remote monitoring systems.The application of advanced sensing devices and artificial intelligence algorithms offers a potential solution to this problem.In this thesis,we design a framework of remote monitoring system for the elderly by innovatively integrating advanced technical means such as Io T gateway,edge computing and cloud computing based on the study of existing remote monitoring technologies.A health prediction model is trained using elderly vital sign data,and a wavelet neural network(WNN)is applied to the model training process to propose a WNN-based health status prediction model for the elderly.Based on this,an improved adaptive nonlinear particle swarm algorithm(NLAPSO)is used to optimize the prediction model to form a NLAPSOWNN-based health prediction model.The specific contents and innovations of this thesis are as follows.(1)Structural design of an Io T-based remote monitoring system.The framework of remote monitoring system is built based on the three-layer standard system of Io T sensing layer,transmission layer and application layer.We decompose,pool and couple the subsystems of "people","things" and "rings" involved in the monitoring of the elderly to form a pool of equipment containing a large number of devices and a pool of data containing a large amount of data.Upgrade wards,nursing homes,and family housing as edge nodes by adding edge servers.The edge nodes are unified and managed by the cloud to realize the "cloud-side collaboration" of the system.The architecture is innovative in four aspects: decomposition,heterogeneous fusion,coupling,and pooling,aiming at efficient data collection,local processing of data through edge computing,and material deployment through "edge-side collaboration",providing a reasonable solution for elderly guardianship from the structural level.(2)WNN-based health status prediction model design for the elderly.The EWS early warning score method in clinical medicine is applied to the pre-processing process of WNN input data,converting elderly vital signs data into score data and determining the health status level according to the total score interval.The WNN neural network model was trained using elderly vital signs data,and the prediction accuracy reached90.231%,which is 4.878% higher than the traditional BP neural network model.(3)NLAPSO-WNN based predictive model design for health status of the elderly.The PSO algorithm is improved by adding an adaptive factor in the form of a nonlinear component with.The initial WNN weights and thresholds are further optimized with the improved NLAPSO algorithm.The experimental results show that the prediction accuracy of NLAPSO-WNN reaches 93.497%,which is 8.144%,5.327%,and 3.266% higher than the traditional BP,PSO-BP,and WNN models,respectively.The system design scheme and health condition prediction model proposed in this paper are expected to provide a solution for remote monitoring data transmission for the elderly and improve the accuracy of health condition prediction for the elderly,thus providing an efficient,reliable and comfortable remote monitoring service.