Paul Mathilde C, Goutard Flavie L, Roulleau Floriane, Holl Davun, Thanapongtharm Weerapong, Roger François L, Tran Annelise
CIRAD, UPR AGIRs, F-34398, Montpellier, France.
IHAP, Université de Toulouse, INRA, ENVT, Toulouse, France.
Sci Rep. 2016 Aug 4;6:31096. doi: 10.1038/srep31096.
The Highly Pathogenic Avian Influenza H5N1 (HPAI) virus is now considered endemic in several Asian countries. In Cambodia, the virus has been circulating in the poultry population since 2004, with a dramatic effect on farmers' livelihoods and public health. In Thailand, surveillance and control are still important to prevent any new H5N1 incursion. Risk mapping can contribute effectively to disease surveillance and control systems, but is a very challenging task in the absence of reliable disease data. In this work, we used spatial multicriteria decision analysis (MCDA) to produce risk maps for HPAI H5N1 in poultry. We aimed to i) evaluate the performance of the MCDA approach to predict areas suitable for H5N1 based on a dataset from Thailand, comparing the predictive capacities of two sources of a priori knowledge (literature and experts), and ii) apply the best method to produce a risk map for H5N1 in poultry in Cambodia. Our results showed that the expert-based model had a very high predictive capacity in Thailand (AUC = 0.97). Applied in Cambodia, MCDA mapping made it possible to identify hotspots suitable for HPAI H5N1 in the Tonlé Sap watershed, around the cities of Battambang and Kampong Cham, and along the Vietnamese border.
高致病性禽流感H5N1病毒目前在几个亚洲国家被认为是地方性流行病毒。在柬埔寨，该病毒自2004年以来一直在家禽群体中传播，对农民生计和公共卫生产生了巨大影响。在泰国，监测和防控对于防止H5N1病毒的任何新传入仍然至关重要。风险地图绘制可以有效地为疾病监测和防控系统做出贡献，但在缺乏可靠疾病数据的情况下是一项极具挑战性的任务。在这项工作中，我们使用空间多标准决策分析（MCDA）来绘制家禽中H5N1高致病性禽流感的风险地图。我们旨在：i）基于来自泰国的数据集，评估MCDA方法预测适合H5N1的区域的性能，比较两种先验知识来源（文献和专家）的预测能力；ii）应用最佳方法为柬埔寨家禽中的H5N1绘制风险地图。我们的结果表明，基于专家的模型在泰国具有非常高的预测能力（AUC = 0.97）。应用于柬埔寨时，MCDA地图绘制使得能够在洞里萨湖流域、马德望市和磅湛市周边以及越南边境沿线识别出适合H5N1高致病性禽流感的热点地区。
