Monitoring Hand, Foot And Mouth Disease Based On Searchengine Query Data

Hand, foot and mouth disease (HFMD) is one of the common infectious disease caused by enterovirus and it is mainly caused by Enterovirus 71 and Coxsackie A16. Like influenza, HIV/AIDS and other diseases which are harmful to the society, the prevention and control of HFMD are absorbing more and more attention from the government and society. The traditional way of monitoring the diseases is to set up a number of sentinel hospitals across the country to count up the cases of a certain kind of disease in a certain period. The monitoring data is then reorganized by statistic before releasing to the society. This approach usually requires a lot of money and time, besides, studies have shown that the official data will only be released for another two weeks after the disease has happened.In order to make up for the shortage of the traditional monitoring methods, digital disease surveillance have been used in recent years. Crowdsourcing data, web mining data and other novel data have been used widely in the monitoring of diseases. Among all the methods of digital disease surveillance, search engine data, represented by Google flu trends, have begun to show its unique value. In 2009, when Google flu trends were firstly put forward, it has absorbed attention widely and become a classical cases of big data application. Many scholars have used Google trends to forecast the trend of economy, unemployment rate, commodity sales, film industry etc. and have achieved some results.Previous research has conducted some application on the search engine data, however, few scholars have evaluated the search engine data from the perspective of space structure.On account of this, this article will collect search engine data in two spatial scale (province and city) and evaluate their spatial distribution and differences from the dimension of geographical space. Besides, this article will use the relevant models to use the search engine data to predict the traditional HFMD data. Finally, the predicting effect of the models will be verified.The full text mainly includes the following work:(1) To begin with, this article analysis the advantages and disadvantages of the traditional monitoring method of HFMD. And monitoring data of HFMD in Guangdong province and its characteristics were analyzed.(2) This article introduce the mainly search engine product like Google trend and Baidu index and list the exploratory search engine products in current internet market. Then the advantages and disadvantages of different products were discussed and we select Baidu index as the mainly research object. Keywords were than obtained from the Baidu index to conduct a visualization in space and time.(3) This article selected 11 keywords, which have a maximum cross correlation more than 0.6, to describe the epidemic trend of HFMD. Based on the analysis of the main social behavior of the guardians of HFMD, the 11 keywords were than classified into three groups which include general keywords, treatment keywords and prevention keywords. After calculating, the composite search index can be got.(4) According to the three kinds of keywords, this article discuss the spatial differences of web searching behavior. The correlation between every kinds of keywords and the real cases in 21 cities were than analyzed. The result turn out that the correlation of between different kinds of keywords and the real cases have aitterent distribution in space.(5) Finally, this article verified the predicting effect of the historical HFMD cases, the collected web search engine data and the combination of historical HFMD cases and the search engine data, the result show that the best predicting result was observed when combining the historical HFMD cases and the search engine data.