Simulation models based on weighted multipartite animal trade networks for the optimized prediction and control of the transmission of classical swine fever

近年来，网络分析已成为表征动物贸易网络的一个有价值的框架。在这里，农场代表网络的节点，而单个农场之间的贸易联系是边缘。除了这些单方网络（即只有一种类型的节点）之外，其他可能的疾病传播途径也可以通过所谓的多方网络（如群体运输、饲料供应）纳入分析。此外，网络分析中还包括边缘权重（如地理距离、运输动物数量）。从而验证是否给出了更准确的疾病传播风险预测。基于边权和不同节点类型构建了不同的网络变化，并通过灵敏度分析进行了比较。由此，可以确定影响网络分析结果的主要变量。因此，对原研究项目的分析得以继续，开放的研究问题得以全面解决。由于最近的研究表明，大多数参数或算法目前仅适用于无向网络（即边缘方向被忽略），因此，本研究项目的后续目标是将这些参数适用于有向多部网络。特别是对于动物贸易网络，以及在石勒苏益格-荷尔斯泰因州（Vermarktungsgesellschaft f<e:1> r Zucht- und Nutzvieh eG）贸易网络的情况下，边缘方向对于预测疾病传播和实施适当的控制措施非常重要。因此，可以将生猪交易网络提供的基本信息纳入分析。下一步，基于这些加权多方网络建立仿真模型，用于预测猪瘟病毒的传播和实施控制策略。多种传播路径以及边缘权重的整合产生疾病传播的真实图像，然后将其与网络分析的结果进行比较。因此，目前的仿真模型不同于经典的仿真模型。在这里，只打算在节点之间建立随机连接。在仿真模型中，分别实施了基于eu -立法、网络和中心性参数的不同控制措施，并验证了其有效性。将模拟研究结果与加权多方网络分析结果进行比较，可以确定适当和可靠的数据基础，以优化疾病传播预测，并为在流行病情况下制定适当的控制策略提供见解。因此，有可能分解潜在的贸易网络并中断感染链。

In recent years, network analysis has become a valuable framework for the characterisation of animal trade networks. Here, the farms represent the nodes of the network, whereas the trade contacts between the single farms are the edges. Besides these monopartite networks (i.e. only one type of nodes), other possible disease transmission paths can be included in the analysis with so-called multipartite networks (e.g. group transports, feed supply). Moreover, edge weights (e.g. geographical distance, number of transported animals) are included in the network analysis. Therewith, it can be validated whether a more accurate prediction of the risk for disease transmission is given.Based on the edge weights and the various node types different network variations are built and compared by a sensitivity analysis. Thereby, the main influencing variables on the outcome of the network analysis can be identified. Thereby, the analyses of the original research Project are continued and open research questions can be comprehensively addressed.Due to the fact that recent studies showed that the majority of the parameters or algorithms are currently only available for undirected networks (i.e. the edge direction is neglected), a follow-up aim of this research project is, thus, to adapt these parameters to directed multipartite networks. Especially for animal trade networks and with this also in the case of the trade network of the ZNVG (Vermarktungsgesellschaft für Zucht- und Nutzvieh eG) Schleswig-Holstein the edge direction is important for the prediction of disease transmission and the implementation of appropriate control measures. Therewith, essential information which is provided by the pig trade network can be included in the analysis.In the next step, simulation models based on these weighted multipartite networks are established which allow the prediction of disease transmission and the implementation of control strategies for the classical swine fever virus. The integration of the versatile transmission paths as well as edge weights results in a realistic image of the disease transmission which is then compared to the outcome of the network analysis. Thus, the present simulation models differ from the classical simulation models. Here, only random connections between the nodes are intended. Furthermore, in the present simulation models, different control measures based on EU-legislative as well as on network and centrality parameters are implemented and their efficiency is verified.The comparison of the results from the simulation study and the weighted multipartite network analysis allows the determination of an appropriate and reliable data basis for an optimized prediction of disease transmission and provides insights in the development of suitable control strategies in the case of an epidemic. Thus, it becomes possible to decompose the underlying trade network and to interrupt the chain of infection.