Weather-based predictive models for Dengue fever epidemics in Vietnam

• Main Authors: Thi-Binh-Minh Nguyen, 阮氏萍明
• Other Authors: Yu-Chun Wang
• Format: Others
• Language: en_US
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/80667477580092805522

碩士 === 中原大學 === 土木工程研究所 === 101 === Dengue fever is one of the most dangerous infectious diseases in Vietnam. This study aims to analyze the associations between climatic factors and socio-economics variables and reported dengue fever cases using temporal and spatial models in Vietnam from 2000 to 2007. This study first used Autoregressive Integrated Moving Average model (ARIMA) to determine autocorrelations between denge fever incidence and five weather factors, such as rainfall, mean temperature, maximum temperature, minimum temperature and relative humidity in eight areas in Vietnam. We further used ARIMA by including weather variables, Population Index and Monthly Population Index to forecast number of dengue cases in Vietnam. This study identified maximum temperature has the most significant influence on dengue epidemics in almost all areas (6 over 8 areas when running model between single variable and number of dengue cases) and it can be used effectively to predict dengue fever epidemics. This study further conducted a spatial analysis by Geographically Weighted Regression (GWR) to model the associations between socio-economics factors and reported dengue fever cases from 2006 to 2007 in whole Vietnam. The study was carried out in three levels. First model analyzed the associations between socio-economics covariates and dengue fever epidemics in eight areas in Vietnam (regional level; eight areas: Northeast, Northwest,Red River Delta, North Central, South Central, Highlands, Southeast, Nine Dragon Delta). Second model determined the relationships between social factors and dengue epidemics in 64 provinces in Vietnam (provincial level). Last model presented effects of socio-economics conditions on reported dengue cases in provinces in each area of Vietnam. The strength of regression is explained by determinant coefficient (R-Square value) and significant level is determined by Moran’s Index through spatial autocorrelation test (or Moran’s Index test). We found influence of socio-economics variables on dengue epidemics in each regional area was clearer than in whole Vietnam (in national scale, we obtained dispersed/clustered patterns in many models and had R-square value < 0.5). Population and population density were two factors that had the strongest impact on dengue fever outbreaks. While Northern part of Vietnam witnessed the absolute victory of population density, Southern part of Vietnam saw the power of population.