School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney 2052, Australia.
Prev Vet Med. 2011 Dec 1;102(3):218-29. doi: 10.1016/j.prevetmed.2011.07.009.
The definition of the spatial relatedness between infectious and susceptible animal groups is a fundamental component of spatio-temporal modelling of disease outbreaks. A common neighbourhood definition for disease spread in wild and feral animal populations is the distance between the centroids of neighbouring group home ranges. This distance can be used to define neighbourhood interactions, and also to describe the probability of successful disease transmission. Key limitations of this approach are (1) that a susceptible neighbour of an infectious group with an overlapping home range - but whose centroid lies outside the home range of an infectious group - will not be considered for disease transmission, and (2) the degree of overlap between the home ranges is not taken into account for those groups with centroids inside the infectious home range. We assessed the impact of both distance-based and range overlap methods of disease transmission on model-predicted disease spread. Range overlap was calculated using home ranges modelled as circles. We used the Sirca geographic automata model, with the population data from a nine-county study area in Texas that we have previously described. For each method we applied 100 model repetitions, each of 100 time steps, to 30 index locations. The results show that the rate of disease spread for the range-overlap method is clearly less than the distance-based method, with median outbreaks modelled using the latter being 1.4-1.45 times larger. However, the two methods show similar overall trends in the area infected, and the range-overlap median (48 and 120 for cattle and pigs, respectively) falls within the 5th-95th percentile range of the distance-based method (0-96 and 0-252 for cattle and pigs, respectively). These differences can be attributed to the calculation of the interaction probabilities in the two methods, with overlap weights generally resulting in lower interaction probabilities. The definition of spatial neighbourhood has important implications for models used in decision-support systems for disease preparedness and response. This research presents a first step towards more realistic representations that could be used in spatio-temporal models of disease outbreaks.
传染病动物群体与易感染动物群体之间空间关联性的定义，是疾病爆发时空建模的一个基本组成部分。在野生动物和野生种群中，疾病传播的常见邻域定义是相邻群体栖息地中心之间的距离。这个距离可以用来定义邻域相互作用，也可以用来描述疾病传播成功的概率。这种方法的主要局限性在于：(1)具有重叠栖息地的传染病群体的易感邻居，如果其中心位于传染病群体的栖息地之外，则不会被认为有疾病传播的可能；(2)对于那些中心位于传染病栖息地内的群体，没有考虑到栖息地重叠的程度。我们评估了基于距离和基于范围重叠的疾病传播方法对模型预测疾病传播的影响。范围重叠是使用模拟为圆形的栖息地范围计算的。我们使用了 Sirca 地理自动机模型，该模型使用了我们之前描述的德克萨斯州九个县研究区域的人口数据。对于每种方法，我们应用了 100 次模型重复，每次重复 100 个时间步长，针对 30 个索引位置。结果表明，基于范围重叠的方法的疾病传播速度明显低于基于距离的方法，使用后者模拟的中位数爆发规模大 1.4-1.45 倍。然而，这两种方法在感染区域方面表现出相似的总体趋势，基于范围重叠的中位数（牛为 48 和 120，猪为 48 和 120）落在基于距离的方法的第 5-95 百分位范围内（牛为 0-96，猪为 0-252）。这些差异可以归因于两种方法中相互作用概率的计算，重叠权重通常会导致较低的相互作用概率。空间邻域的定义对用于疾病准备和应对决策支持系统的模型具有重要影响。本研究朝着更现实的表示方法迈出了第一步，这种方法可以用于疾病爆发的时空模型。
