新疆巴州地区布鲁氏菌病模型的分析与仿真

发布时间：2018-08-29 12:24

【摘要】：目的:为了探讨传染病动力学模型、时间序列模型及Richards模型在新疆巴州地区布鲁氏菌病研究与应用中的可行性,根据巴州地区布鲁氏菌病的实际发病情况,建立相符的布鲁氏菌病传播流行的动态模型,掌握巴州地区布鲁氏菌病的总体流行趋势,为布鲁氏菌病的早期预警及控制,奠定科学的理论基础,提供可行的参考依据。方法:首先运用季节指数验证巴州地区布鲁氏菌病的季节性波动规律,构建具有周期性传播率的动力学模型,拟合2010~2014年每季度新发的急性人间布鲁氏菌病数据,MAPE与RMSPE可用于评价模型拟合的效果,并预测未来人间布鲁氏菌病的流行趋势。通过PRCC方法可提取出模型中对人间布鲁氏菌病流行具有统计学意义的参数,计算出布鲁氏菌病传播的基本再生数R0,探究参数改变分别对人间布鲁氏菌病流行趋势及基本再生数R0敏感性分析,提出防控措施。其次时间序列中的ARIMA乘积季节模型可用于研究季节性变化的传染病发病规律,基于巴州地区2005~2014年间每月新发的人间布鲁氏菌病数据,建立ARIMA(P,D,Q)(p,d,q)S模型,最优模型可通过AIC、SBC及AICC最小值原则选取,根据最优模型拟合及预测新发的人间布病流行趋势。最后针对呈暴发型特征的人间布鲁氏菌病数据可运用Richards模型分析。模型拟合巴州地区2013年8月至2014年12月之间累积新发的人间布鲁氏菌病数量,分析暴发期间发病率最高的拐点,有助于研究干预措施相对于拐点的影响,并估计出基本再生数R0。结果:首先根据季节指数分析确定新发的急性人间布鲁氏菌病数量在夏、秋季较高。基于布鲁氏菌病的传播机理分析,构建羊/牛及从羊/牛传播给人的季节性SEIV动力学模型,拟合新发的急性人间布病数量时MAPE=18.07%,RMSPE=20.89%,说明模拟效果理想,预测出大约在2023年的夏季新发的急性人间布鲁氏菌病数量达到最大值15325(95%CI:11920-18242)。估计出基本再生数R0=2.5524(95%CI:2.5129-2.6225),表明疫病还将持续流行,尚不能被消除。参数的敏感性分析可确定,减少羊/牛出生数量,增大已感染布鲁氏菌病羊/牛屠宰率,增加易感染布鲁氏菌病羊/牛免疫接种率,降低免疫接种丢失率为有效控制新发的急性人间布鲁氏菌病流行措施。其次,时间序列模型中原始序列白噪声检验为P0.05,有研究的价值。构建的最优模型为ARIMA(1,1,1)(0,1,2)12,此时AIC=973.12,SBC=987.02,AICC=973.66最小,模型拟合值与实际值误差为MAPE=23.82%,RMSPE=29.64%。根据已知序列预测出2015年6月巴州地区新发的人间布病数量达到最大值91(95%CI:51-131)。最后,Richards模型拟合暴发时新发的人间布鲁氏菌病数量误差为MAPE=6.80%,RMSPE=3.98%,同时估计在2014年6~7月之间人间布鲁氏菌病发病率最高,早于实施各种防控措施。基本再生数R0=1.1207(95%CI:0.6091-1.1379),说明人间布鲁氏菌病尚不能消除。结论:具有周期性传播率的季节性动力学模型、ARIMA乘积季节模型与Richards模型都能较好的模拟出符合各自数据特征的人间布鲁氏菌病流行趋势,可行性较高,为相关工作人员在布鲁氏菌病高发季之前做好预防准备,提供防控参考策略。
[Abstract]:OBJECTIVE: To explore the feasibility of infectious disease dynamics model, time series model and Richards model in the study and application of Brucellosis in Bazhou, Xinjiang. According to the actual incidence of Brucellosis in Bazhou, a corresponding dynamic model of Brucellosis transmission and epidemic was established to master the total Brucellosis in Bazhou. Methods: Firstly, the seasonal fluctuation of Brucellosis in Bazhou was verified by seasonal index, and a dynamic model with periodic transmission rate was constructed to fit the new acute human cases from 2010 to 2014. MAPE and RMSPE can be used to evaluate the fitting effect of the model and predict the epidemic trend of human brucellosis in the future. The parameters with statistical significance for human brucellosis epidemic in the model can be extracted by PRCC method, and the basic reproduction number R0 of brucellosis transmission can be calculated, and the change of parameters can be explored respectively. Secondly, the ARIMA product seasonal model in time series can be used to study the seasonal variation of the incidence of infectious diseases. Based on the new human Brucellosis data from 2005 to 2014, the ARIMA (P, D, Q) (p, d, q) S model is established. The optimal model can be selected by AIC, SBC and AICC minimum principle, and then fitted and predicted the epidemic trend of new human brucellosis according to the optimal model. Results: First, the number of new acute human brucellosis in summer and autumn was determined according to the seasonal index analysis. Based on the analysis of the transmission mechanism of brucellosis, the construction was made. The seasonal SEIV kinetic model of sheep/cattle and human transmission from sheep/cattle fitted the number of new acute human brucellosis with MAPE=18.07% and RMSPE=20.89%, indicating that the simulation effect was satisfactory. The maximum number of new acute human brucellosis in summer of about 2023 was predicted to reach 15325 (95% CI: 11920-18242). 2.5524 (95% CI: 2.5129-2.6225), indicating that the epidemic will continue to prevail and can not be eliminated. Sensitivity analysis of the parameters can be determined to reduce the number of sheep/cattle born, increase the slaughter rate of infected sheep/cattle, increase the immune inoculation rate of susceptible to brucellosis sheep/cattle, and reduce the loss of immune inoculation rate to effectively control new acute people. Secondly, the white noise test of the original sequence in the time series model is P 0.05, which is of research value. The optimal model is ARIMA (1,1,1) (0,1,2) 12. At this time, AIC = 973.12, SBC = 987.02, AICC = 973.66 is the smallest. The error between the fitting value and the actual value of the model is MAPE = 23.82%, RMSPE = 29.64%. The maximum number of new human brucellosis was 91 (95% CI: 51-131) in Bazhou in June, 2004. Finally, the error of fitting the Richards model to the outbreak was MAPE = 6.80%, RMSPE = 3.98%, and the incidence of human brucellosis was estimated to be the highest between June and July, 2014, earlier than the implementation of various control measures. R0 = 1.1207 (95% CI: 0.6091-1.1379), indicating that human brucellosis can not be eliminated. Conclusion: Seasonal dynamics model with periodic transmission rate, ARIMA product seasonal model and Richards model can better simulate the epidemic trend of human brucellosis in accordance with their respective data characteristics, the feasibility is high, for the relevant staff in Brucellosis prepares for prevention before the high season, and provides a reference strategy for prevention and control.
【学位授予单位】：新疆医科大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R516.7

【参考文献】