Application And Research Of Hand-foot-mouth Disease Epidemic Prevention And Control Based On Smart Medical Service

"Healthy China 2030" program clearly put forward:strengthen the prevention and control of major infectious diseases.With the continuous development of machine learning technology,machine learning plays an irreplaceable role in intelligent medical services.More and more algorithm models are used to predict the risk of hand-foot-mouth disease(HFMD).However,there are some problems in the research process of HFMD prediction model:there are many researches on the training of prediction model,but few practical applications by clinicians;There are many researches on algorithms,but few researches on data import methods;Most of the prediction models of HFMD are based on Algorithms and super parameters,but few on other factors;In the early stage of the project,the improvement of the model is paid more attention,and there is a lack of continuous improvement in the use process.The purpose of this study:without the help of engineers,clinicians can independently and conveniently complete the model training of HFMD,and promote the continuous improvement and use of the prediction model.Methods:by introducing business parameters,the doctors were given the right to choose the characteristic attributes such as clinical characteristics,so that the doctors could be more deeply involved in the training of the model;SOAP API and RESTful API are used to realize automatic data processing and transmission;The feature disturbance is formed by the different business parameters selected by doctors.At the same time,a variety of classifiers are introduced to continuously accumulate more hand foot mouth disease prediction models.On this basis,the business parameters of these models are extracted,and the unified data set is used for training and evaluation in the way of grid search.Results:after the introduction of business parameters,the number of doctors using the model to predict HFMD increased from 10.39 times to 19.72 times per day;It takes 358.213 seconds to export 100000 records with 16 column attributes to the CSV file through Oracle,and only 5 seconds for the same data set to import data through API interface;The prediction model based on business parameters is superior to the prediction model based on diagnostic criteria in accuracy,recall,accuracy and F1 evaluation indexes.Conclusion:the introduction of business parameters is beneficial for doctors to actively use HFMD prediction model;Based on SOAP API and RESTful API,the automation degree of data processing and transmission is improved,which is helpful to the training efficiency of HFMD prediction model and the continuous improvement of the model;The prediction effect of HFMD prediction model based on Stacking ensemble learning and business parameters is better.Therefore,the introduction of business parameters and API into the risk prediction system of HFMD based on machine learning has practical significance for the early detection,early diagnosis and early treatment of HFMD.