The use of classification and regress tree to predict incidences of dengue in relation to climatic variables and imported dengue cases in Cairns, Australia

Le résumé fourni par la source

Background: Dengue fever (DF) is a viral disease transmitted by Aedes mosquitoes mainly in tropical and subtropical areas of the world. Recently, there are a number of outbreaks of DF in Cairns, a tropical city in Australia. Aims: To explore the possibility of developing a forecast model of DF based on weekly weather variability and imported DF cases in Cairns, Australia. Method: We obtained data from Queensland Health on the numbers of notified DF cases (acquired locally and overseas) in Cairns for the period 1st January 2000 through 31 December 2009. Data on weather (minimum temperature, maximum temperature, relative humidity and vapour pressure) and population were obtained from the Australian Bureau of Meteorology and the Australian Bureau of Statistics, respectively. A weekly time series classification and regression tree (CART) model was used to evaluate the non-linear relationship between weather variability, imported DF cases and locally acquired DF cases. Cross-validation was used to deal with over-fitting and to identify the optimal tree with respect to its predictive ability. Results: The results indicated that locally acquired DF cases were strongly associated with imported DF cases, average minimum temperature and average maximum temperatures at a lag of 1 - 3 weeks. The CART model showed that the relative risk of locally-acquired DF increased by 16.6-fold (expected weekly incidence rates of DF: 40.15/100,000) when average minimum temperature exceeded 24°C at lags of 1 – 3 weeks, average maximum temperature was under 32°C at lags of 1 – 3 weeks and imported DF case is over 0 at lags of 1 – 3 weeks. Conclusion: These findings may have significant implications for developing a local forecast model to predict DF outbreaks, which can be applied as a decision support tool in planning DF control and prevention programs based on routinely collected data.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.