Research And Implementation Of Family Medical Health Monitoring System

With the passage of time,the degree of aging in China has continued to deepen.According to statistics,there are currently 240 million elderly people in China.It is predicted that the elderly population will exceed 430 million in 2049,accounting for one third of the total population.The health monitoring of the elderly will become the most urgent social problem to be solved.The biggest threat to the health of the elderly is chronic diseases such as cardiovascular disease.According to the National Cardiovascular Disease Center,the number of patients with cardiovascular disease in China is estimated to be 290 million,accounting for more than 40% of the deaths of residents.And accidental falls are also one of the main causes of disability in the elderly.In response to the problem of health monitoring for the elderly,since 90% of China’s old-age care model is characterized by home-based care,home-based care has gradually become the mainstream solution.But the problem with home health monitoring is that there are many types of health monitoring equipment,but they are not compatible with each other.Therefore,the homebased medical health monitoring system proposed in this thesis collects physiological data such as blood pressure,blood glucose,and blood oxygen based on medical equipment that conforms to the ISO / IEEE 11073 standard.And use the HDP protocol to transfer the collected data to the client based on the Android system development.The client can view the management data and synchronize the data to the background server.At the same time,the system also has the functions of fall detection and ECG monitoring.In order to realize the fall detection function,this thesis proposes a fall detection algorithm based on KKT conditional SVM incremental learning,using SVM incremental learning to solve the problem of the accuracy of the model obtained by the experimental data training in the real environment and the user’s use process.The new sample data will cause the problem of too large training samples.The KKT conditional filtering is used to solve the problem of data imbalance in the new samples generated during user use.The UniMiB SHAR fall data set of the algorithm has been verified by the algorithm,and the accuracy of fall detection has reached 97%.Compared with the traditional SVM algorithm,the algorithm improves the training speed without reducing the accuracy and reduces the memory footprint.Finally,this thesis studies the ECG classification algorithm based on DWT and SVM support vector machine to classify the user’s ECG signal into normal heartbeat and arrhythmia heartbeat.In this thesis,based on the MIT-BIH arrhythmia database,discrete wavelet transform is used for denoising,R wave detection and feature vector extraction.From each heartbeat cycle,8 sets of coefficients,4 sets of detail coefficients,and 4 levels of approximate coefficients are obtained by wavelet decomposition.Finally,SVM model training was conducted for 8 different combinations of coefficients.Among them,the 4 groups of detail coefficients performed best,reaching 98%,which can accurately distinguish normal heart beats from arrhythmia heart beats.