Software Design And Implementation Of Sports Health Cloud Platform System

With the development of information technology,intelligent sports equipment and software have appeared in the field of sports to collect human body movement and sign data and display it intuitively,assisting professional sports personnel in planning sports strategies.But for the general public who lacks professional knowledge,intuitive data does not help them make correct exercise plans.Therefore,the effective use of the collected motion and sign data to analyze the user's personal physical condition,and to generate reasonable exercise suggestions has become the research direction in the field of intelligent sports health.Based on the above problems,this article aims to design and implement a cloud platform that can collect user movement and sign data and generate personalized exercise prescriptions for users.The main research content includes the software design and implementation of the cloud platform system and the design and implementation of a core function algorithm that can generate personalized exercise prescriptions.The main work here is as follows:In the software design and implementation of sports health cloud platform system,this article designs the overall structure of the cloud platform by analyzing requirements,modular design and other methods,and then selects the technical solution and adopts the combination of Springboot + Nginx + tomcat + Mysql + Redis.Functionally,the cloud platform receives and stores user data,generates personalized sports prescriptions,historical data displays,sports prescriptions display and other functions;in terms of performance,it can meet the high-traffic stable access requirements of 4.8 million pv / day.In the design and implementation of the core function exercise prescription generation algorithm,By the method of consulting literatures,this paper compares the performance of random forest with common classification algorithms such as ANN,SVM,KNN,and finds that random forest has the highest accuracy and the best ROC and F1-score performance index.,Then use the body measurement data set used in this article and perform exercise prescription generation performance tests on the corresponding exercise prescription labels.The results show that the random forest has the best accuracy compared to the above commonly used classification algorithms,Recall and F1-score values.Therefore,In this paper,random forest is selected as the algorithm for function realization.Then,in view of the shortcomings of the random forest algorithm in the voting mechanism that ignore the difference in sub-tree classification ability and the performance of non-equilibrium sets,there is still room for improvement.In this paper,the AUC value obtained through literature reviewing,theoretical demonstration and experimental testing can be used as the voting weight parameter improves the conclusion of the random forest algorithm's voting mechanism.After testing,this method can improve the performance of various indicators such as accuracy,precision,and recall rate of the algorithm in the balance set,with a maximum increase of 7.6%;it can also improve the algorithm In the unbalanced concentration of recall performance,specificity,G-mean and other index performance,the specificity of the minority classification performance index can be increased up to 33.3%.Finally,through the test experiment of the exercise prescription generation,the results show that the improved random forest algorithm and the original random forest,the accuracy-weighted random forest compares the performance of the prescription generation,and the improved algorithm has a comprehensive improvement in accuracy of respectively 9.03%,2.71%;the recall rate increased by 7.31% and 2.69%,respectively.