Temporal And Spatial Clustering And The Incidence Trend Prediction Of Hand Foot Mouth Disease In Changsha City

A total of 100199 cases of hand, foot and mouth disease reported in ChangSha city from 2011 to 2014, the total number of patients, severe cases and death toll present high incidence characteristics in even-numbered years. In order to better understand the HFMD distribution characteristics of Changsha city and to effectively allocate medical and health resources, government have goals and focus on the prevention and control of diseases. This article reveals time, population and spatial distribution characteristics from the epidemiology of HFMD during 2011-2014 in Changsha city. Spatial-temporal analysis of HFMD incidence by month and township as a unit. Finally, to predict the onset of hand, foot and mouth disease, using the combination of the average number of HFMD and the relevance of average meteorological factors within a week and establishing the mathematical model. The results of this study summarized as follows.(1) Epidemiological characteristics of HFMD.During 2011 to 2014, the male patients in a total of 60485 cases. The cases have focused on 1-4 years old, most affected children were scattered children and accounted for 76.03%. The incidence of HFMD peak is different and present state of double peak, and cases distributed all townships in Changsha.(2) Space and temporal clustering analysis. Spatial-temporal analysis of HFMD incidence by month and township was conducted with SatScan software. We used MapGIS software to establish the database and display cases in time and space gathering area. Purely temporal and purely spatial scan statistics indicated Class 1 clustering occurred in Xingsha Xianglong Quantang and Xiaohe township from April to July during study period. However, HFMD cases focus from April to June in 2014, and mainly concentrated in the central and southeastern region by spatial-temporal scan statistics in Changsha during 2011-2014.(3) Correlation analysis. HFMD cases and meteorological factors were collected weekly from 2010 to 2014 in Changsha city.Using the Generalized estimating equations (GEE) with Negative binomial regression to filter the average of temperature, pressure, relative humidity, wind speed, lowest temperature, horizontal visibility and precipitation within a week by SAS software. The article finally showed that five meteorological factors had statistical significance, specifically, the weekly average temperature, weekly average pressure, weekly average visibility and so on.(4) The Back Propagation Neutral Network (BPNN) prediction. In order to predict trend of HFMD, the 260 weeks of average HFMD cases and weeks average meteorological data distribution into the proportion of 8:2. Among them,1-182 weeks data as the training sample data,2-183 weeks as the training target,182-259 weeks data as the test sample data, and 183-260 weeks as the objective of the test data. The results showed that the BPNN model has better prediction accuracy, it can be used for HFMD incidence trend prediction research.