Diabetes-assisted Medical Processing System Based On Big Data Technology

Diabetes is one of the major chronic non-communicable diseases in the world.Its complications can cause serious harm to the patient's body and even threaten life.The core treatment of diabete is glycemic control,but this process consumes a lot of resources,which is a burden on patients' families and the society.Although diabetes is currently incurable,but diabetic complications can be effectively prevented or delayed through healthy eating,reasonable exercise,positive attitude and timely medication.In recent years,with the development of new technologies such as big data,cloud computing and deep learning,the integration of various disciplines and information technology can play a greater role in the prevention and treatment of diabetes.With the help of big data technology to analyze the collected physiological data of diabetic patients,using modern intelligent devices to greatly simplify the user's data input steps can give patients a reasonable life advice and reduce the user's burden.Based on the above ideas,this paper proposes a diabetes data processing framework based on big data technology.The framework is divided into two parts,front end and back end.Supported by big data technology,the backend leverages a Lambda architecture that combines MongoDB databases,Kafka Message Queuing and the Spark ecosystem for data storage,mass off-line batch and live stream processing.The front end is a mobile APP,responsible for the collection and processing of user data.APP automatically synchronizes the data with the wearable device via Bluetooth,front-end and back-end communication through the network to complete the interaction.The key issue for automated data acquisition is data synchronization for wearable devices.This paper analyzes the workflow of Bluetooth protocol,using Android development tools to experiment,the experimental results show that we can obtain wearable device data through technical means.Finally,this paper presents a method of predicting calories from food images based on depth learning technology.After a user takes a photo of the food item,the recognition model identifies the location and type of ingredients in the photo,predicts the weight by the weight prediction method,and finally calculates the calories.This paper collected food image data set training SSD object detection model as a food recognition model,the data set is subjected to polynomial regression to fit the food weight prediction function,by querying the US Department of Agriculture national nutrition database to get the food's heat density,and finally the weight By multiplying the caloric density,caloric count can be calculated.Experimental results show that the accuracy of the recognition model reaches 94% and the thermal prediction error is about 10%,which has a certain practical value.