Application Of Deep Learning In Human Health Monitoring System Based On Body Surface Temperature

With the increasing concern for health status,health monitoring equipment has also developed vigorously.The current mainstream products on the market can only monitor and display the physiological data of body.Most health monitoring products can not analyze health status.A small part of products with reminder function need to manually set the threshold.Such products are prone to false alarms in different health status and the analysis of human health status has serious defects based on only one kind of data.In view of the defects of the existing products,this paper selects the body surface temperature as the physiological characteristic which can best reflect the health state of human body and is closely related to other physiological characteristics.This paper discusses the application of deep learning in human health system based on body surface temperature.The system accomplishes data acquisition and transmission by means of wireless communication with many kinds of sensors,analyzes the collected data intelligently using the deep neural network constructed and trained,and designs a new threshold evaluation method.It mainly realizes the monitoring function of the user’s health status and the recording function of the health status change.The experiment of the health monitoring system meets the basic requirements of the medical system,and meets the requirements of this paper.The main work of this paper is as follows:(1)The overall scheme of the hardware and software of the system is designed and the actual objects are made for testing.The hardware platform can accomplish the continuous collection and wireless transmission of the required physical signs and environmental data.The acquisition precision is up to the common standard of the existing health detection equipment.The software platform displays the user’s knowledge and visualizes the ordered patterns from data mining.(2)A model of mental and physiological health status recognition algorithm based on combined neural network is designed.RNN is widely used in time series.In this paper,the RNN is used as the representation network,the logic unit uses the short and long term memory model and adds a convolutional neural network layer with a random dropout layer before the LSTM unit,taking into full account the data characteristics of multi-dimensional physical sign data to improve the accuracy.By training on professionally labeled data sets,it is possible to distinguish temperature changes in a variety of daily human conditions from those caused by physical or mental stress.(3)Based on the above work,a health monitoring and evaluation method is designed,which can avoid the false report caused by the effects of sports,happiness and sadness on body temperature through the intelligent analysis and prediction of the physical and mental health of human body.Combined with the prediction of body temperature based on environmental variables,compared with the traditional threshold method,the accuracy and reliability of health monitoring system are effectively improved,and the situation of false alarm is effectively reduced,it also provides complete and more accurate auxiliary information for medical staff,and improves the effect and comprehensiveness of the monitoring system.