Research And Design Of Embedded Health Information System

At present,most health information monitoring devices on the market are used to monitor a certain aspect of health abnormality.In order to fully monitor the health status of patients help in hospitals,nursing homes and such kind of places,it has designed an embedded health information processing system which is superior to previous mobile monitoring devices.In the article,the system is fixedly installed on the indoor ceiling or wall,and a variety of algorithms are used to achieve healthy real-time monitoring.Firstly,it has introduced all the functions and the whole framework of the system.Secondly,the technical schemes of each functional module are proposed.The system uses neural network algorithm to detect health anomalies of the received the health index data.MLX90614 sensor is used to achieve the noncontact temperature measurement with an accuracy of ±0.5?.Based on the UWB positioning technology,and TOA algorithm,it can realize the accurate location in the building of the patient precisely to centimeter level.Besides,the fall detection algorithm is used to realize the fall detection of the human body through the extraction of the characteristics of the moving target in the video,and the face recognition method is used to identify the user's identity.Thirdly,A face recognition method based on capsule network algorithm is designed in this paper.The capsule network algorithm uses the active vector to represent the existing face features,which reduces the error rate caused by the representation of scalar.And it uses the dynamic routing between Capsule to train the weight parameters of the capsule network.The capsule network structure of face recognition is composed of input layer,Convolution layer,Primarycaps layer,Digitcaps layer,three full connection layers and output layer respectively.The structure could classify and recognize the input image of face.We build face data and set parameters to train the structure.As the result of training,the accuracy of face recognition based on the capsule network algorithm reaches 98%.Finally,the TensorFlow platform is built to realize face recognition.With the program in Python3.6 language,and the code on the Python IDE,the face recognition experiment of the single face,the two faces,the three faces and the four faces are finished.The experimental results show that the identification accuracy of single person and double person is relatively high and the accuracy of more than three faces need further optimization of capsule network.The identification function based on face helps medical staff to understand the patient's medical history and health status in advance and quickly,so that the health information processing system can work more effectively.