A dataset of a proxy variable for the poultry trade flows in China

Abstract

Understanding the intercity poultry trading network is crucial for assessing the risk of avian influenza prevalence. Unfortunately, the poultry trading network in China has rarely been described. Here, with a modified radiation model, we obtain values for a proxy variable for poultry trade flows among 318 prefecture-level cities in China in 2015 utilizing the product capacity and demand quantity of poultry of the cities. The results are validated by comparing the proxy variable values with the trade volumes investigated in the literature, and it is found that the modified radiation model can accurately predict the main poultry trade flows among cities. This is the first dataset on China’s poultry trade pattern, and it can be used to analyze the production and consumption structure of poultry in prefecture-level cities within China. The dataset can be a tool for avian influenza epidemic risk assessment as well as a basis to develop prevention and control measures during an epidemic.

Simulating contact networks for livestock disease epidemiology: a systematic review

“…[ 66 ]) 2016 mechanistic ABM structure and transmission yes cattle Cameroon foot and mouth disease geo-locations livestock movement dynamic network survey 17 (Kong et al . [ 67 ]) 2022 mechanistic radiation model limited data no poultry China non-specific geo-locations livestock movement static emergent 18 (Valdes-Donoso et al . [ 68 ]) 2017 machine learning RF limited data, network-generating processes no pigs USA porcine reproductive and respiratory syndrome livestock holdings, markets livestock movement static network survey 19 (Nicolas et al .…”

“…Kong et al . [ 67 ] adapted the radiation model to predict country-scale poultry flows in China, with poultry population representing supply ( m i ), and human populations representing demand ( n j ) and ‘intervening’ demand ( s ij ).…”

“…Approaches for external validation included splitting the data into training and validation sets (e.g. [ 68 ]), or through comparison with different datasets [ 15 , 56 , 65 , 67 ], such as for different time points [ 74 , 85 , 87 ]. A single GM was externally validated by assessing whether observed changes in livestock movements resulting from demand changes (i.e.…”

“…Other models were internally validated at the level of the dyad (n = 6). Examples of approaches here included computing the predictive accuracy of binary or weighted edges based on, respectively, the area under the receiver operating characteristic curve, or correlation coefficients ( [69] GM, [25] other statistical, [68] RF; Kong et al [67] radiation model). Distributions of observed and predicted geographical distances between connected dyads were also sometimes compared ( [86] other statistical, [68] RF, [56] ABM).…”