基于ZINB模型和气象因素的禽霍乱发病预测

为了评估气象因素对禽霍乱流行的影响,实现对禽霍乱疾病的预警与可视化管理。该研究设计了基于季节、时间分析方法与零膨胀负二项式(zero-inflated negative binomial,ZINB)模型的禽霍乱风险分析方法。该方法利用季节性分析、时间序列分析方法,对中国禽霍乱的分布特征进行描述;通过构建ZINB模型,对气象因素与禽霍乱发生之间的关系进行评估。最终,利用模型分析结果与网络地理信息(Web GIS)技术建立禽霍乱监测预警系统。结果表明,中国禽霍乱发病存在明显的季节性并受上月发病数的影响,同时,较高的气温、空气湿度、较小的风速是促进禽霍乱发病的气象条件。利用该研究中的方法结合气象数据,对禽霍乱发病进行预测,预测值与实际值拟合结果ICC(组内相关系数)取值0.802,R2值为0.7853,表明模型预测精度较好,系统各功能模块运行正常。利用Web GIS技术与数学模型对禽霍乱进行监测预警,可以提高家禽养殖管理水平,降低发病风险,减少因动物疫病造成的经济损失。

Prediction of Avian pasteurellosis based on ZINB model and meteorological factors

Abstract: According to reliable data, the epidemic of poultry infectious diseases causes billions of economic losses every year in China. It not only seriously threatened the healthy development of the poultry industry, but also had a serious impact on the food security and human health. Compared with the developed countries, the problem of the poultry disease prevention in China is that farmers always ignore the importance of comprehensive management and control, and, tend to excessive reliance on vaccines and antibiotics for the infectious diseases. Experience has shown that without the daily management, only relying on antibiotics and vaccines can''t effectively control the spread of infectious disease. This is because that although vaccines can improve the immunity of animal, they can''t kill pathogenic microorganism which have already existed in animal and farm. At the same time, in the face of new infectious diseases due to continuous mutation of pathogenic microorganism, the development of a new novel vaccines always have a lag period. In this case, evaluation of the risk factors which can affect the spread of the disease will help to develop and to strengthen chicken farm management, resulting effective disease control measure. As for antibiotics, the problem of drug-resistant has produced serious influence on the treatment of related diseases. In addition, antibiotics in animal waste ended in environment also have been reported to cause problems in ecosystems and to human health. The purpose of this paper is to study the effect of meteorological factors on the prevalence of Avian pasteuretlosis by using a variety of statistical methods and mathematical model. Data for poultry infectious diseases between Jan., 2006 to Dec., 2016 were obtained from China National veterinary. Provincial meteorological data were obtained from China State Meteorological Administration. By using seasonal analysis, time series analysis method and the zero-inflated negative binomial model, trend analysis and evaluation model of Avian pasteuretlosis have been conducted. This study included: 1) By using seasonal analysis and time series analysis, we studied the epidemic tendency of Avian pasteuretlosis. From the results, we found that the Avian pasteuretlosis, which have been included in the "Official Veterinary Bulletin", was influenced by the seasonal factors and the number of disease cases. 2) By using Zero-inflated negative binomial model, disease data of Avian pasteuretlosis in months and the provincial level meteorological data in every month of the whole country during research period, we reviewed the relationship between meteorological factors and Avian pasteuretlosis cases. Results indicated that there was a high possibility of Avian pasteuretlosis when the weather conditions with high temperature and air humidity and low wind speed. In summary, we studied the epidemic trend of Avian pasteuretlosis and the effect of meteorological factors on this poultry disease. The study offered a reliable theory for setting up the early warning system of Avian pasteuretlosis, and it also provided a reference to the control of other poultry infectious diseases. In further research work, more risk factors, such as the economy, traffic and the migration of migratory birds, should be considered in the analysis to improve the accuracy of the model prediction.