武汉市江汉区细菌性痢疾发病率的ARIMA模型预测应用

目的：探讨A RIM A模型在武汉市江汉区细菌性痢疾月发病率预测应用的可行性,为该区细菌性痢疾的防控提供科学参考依据。方法使用SAS9．2软件对2005～2013年该区菌痢月发病资料拟合ARIMA模型,利用建立的模型对2014年1～6月菌痢月发病率资料进行预测和效果评价。结果建立ARIMA（0,1,1）X（0,1,1）12模型拟合效果较好,预测2014年上半年疫情将呈缓慢上升趋势,预测值与实际值拟合趋势基本一致。结论 A RIM A模型可以作为该区菌痢月发病水平的短期预测模型。     

ARIMA模型和BP神经网络模型在我国乙型肝炎发病预测中的应用

目的探讨适合全国乙肝发病率的预测模型,为乙肝预测预警系统提供参考。方法应用2004-2012年全国乙肝月发病率数据,分别建立ARIMA模型和BP神经网络模型,利用建立的模型预测2013年1-12月乙肝发病率,采用实际发病率验证与比较两种模型的预测效果,评价指标为平均绝对误差(MAE)、平均绝对误差率(MER)和非线性相关系数(RNL)。结果全国2004-2013年乙肝月发病率在2.79/10万~9.44/10万间波动,序列具有明显的长期趋势。建立的乘积ARIMA(0,1,1)(0,1,1)12模型预测的MAE、MER、RNL分别为0.445、0.065、0.909,BP神经网络模型分别为0.635、0.093、0.872。ARIMA模型预测的平均绝对误差和平均绝对误差率要低于BP神经网络模型(△MAE=-0.190,△MER=-0.028),非线性相关系数要高于BP神经网络模型(△RNL=0.037)。结论 ARIMA模型和BP神经网络模型均适用于我国乙肝发病率的预测,且前者的预测效能和非线性拟合能力略优于后者。     

ARIMA模型、BP神经网络及其组合模型在卫生政策评估中的实证比较:以公立医院价格改革为例

目的:探索不同反向事实构建方法对医院财务数据预测的效率,以期对政策进行更有效的评估。方法:借助R软件,用南京市公立医院A在2011—2016年的药品收入、医疗服务收入建立测试数据集,分别用ARIMA模型、BP神经网络、ARIMA+BP组合模型进行预测并与实际拟合,并比较改革前后补偿率。结果:三个模型对药品收入的均方根误差分别为692.82、501.44、380.80,医疗服务收入的均方根误差分别为184.04、215.63、168.65,组合模型预测效率更高。用组合模型计算改革后A医院药品收入净损失为12 044.03万元,医疗服务收入净增长为18 532.60万元,为药品收入损失的153.87%。结论:医院财务数据因其线性与非线性的组合特征,使用组合预测模型的预测效果最佳。但在实际应用中,ARIMA模型操作简单,与组合模型预测趋势也较为一致,在实际卫生政策评估中也推荐使用。     

The impact of immunization programs on 10 vaccine preventable diseases in Italy: 1900–2015

BackgroundVaccination has determined a dramatic decline in morbidity and mortality from infectious diseases over the last century. However, low perceived risk of the infectious threat and increased concern about vaccines’ safety led to a reduction in vaccine coverage, with increased risk of disease outbreaks.MethodsAnnual surveillance data of nationally communicable infectious diseases in Italy between 1900 and 2015 were used to derive trends in morbidity and mortality rates before and after vaccine introduction, focusing particularly on the effect of vaccination programs. Autoregressive integrated moving average models were applied to ten vaccine-preventable diseases: diphtheria, tetanus, poliomyelitis, hepatitis B, pertussis, measles, mumps, rubella, chickenpox, and invasive meningococcal disease. Results of these models referring to data before the immunization programs were projected on the vaccination period to estimate expected cases. The difference between observed and projected cases provided estimates of cases avoided by vaccination.ResultsThe temporal trend for each disease started with high incidence rates, followed by a period of persisting reduction. After vaccine introduction, and particularly after the recommendation for universal use among children, the current rates were much lower than those forecasted without vaccination, both in the whole population and among the 0-to-4 year olds, which is, generally, the most susceptible age class. Assuming that the difference between incidence rates before and after vaccination programs was attributable only to vaccine, more than 4 million cases were prevented, and nearly 35% of them among children in the early years of life. Diphtheria was the disease with the highest number of prevented cases, followed by mumps, chickenpox and measles.ConclusionsUniversal vaccination programs represent the most effective prevention tool against infectious diseases, having a major impact on human health. Health authorities should make any effort to strengthen public confidence in vaccines, highlighting scientific evidence of vaccination benefits.     

Bayesian age–period–cohort models with versatile interactions and long‐term predictions: mortality and population in Finland 1878–2050

Age–period–cohort (APC) models are widely used for studying time trends of disease incidence or mortality. Model identifiability has become less of a problem with Bayesian APC models. These models are usually based on random walk (RW1, RW2) smoothing priors. For long and complex time series and for long predicted periods, these models as such may not be adequate. We present two extensions for the APC models. First, we introduce flexible interactions between the age, period and cohort effects based on a two‐dimensional conditional autoregressive smoothing prior on the age/period plane. Our second extension uses autoregressive integrated (ARI) models to provide reasonable long‐term predictions. To illustrate the utility of our model framework, we provide stochastic predictions for the Finnish male and female population, in 2010–2050. For that, we first study and forecast all‐cause male and female mortality in Finland, 1878–2050, showing that using an interaction term is needed for fitting and interpreting the observed data. We then provide population predictions using a cohort component model, which also requires predictions for fertility and migration. As our main conclusion, ARI models have better properties for predictions than the simple RW models do, but mixing these prediction models with RW1 or RW2 smoothing priors for observed periods leads to a model that is not fully consistent. Further research with our model framework will concentrate on using a more consistent model for smoothing and prediction, such as autoregressive integrated moving average models with state‐space methods or Gaussian process priors. Copyright © 2013 John Wiley & Sons, Ltd.     

应用ARIMA模型预测成都市新都区麻疹的发病趋势

[目的]预测成都市新都区麻疹的发病趋势,为制定防治措施提供参考依据。[方法]以1985～2008年成都市新都区麻疹年发病率作为历史序列,建立自回归移动平均(Autoregressive Integrated Moving Average,ARIMA)模型。[结果]采用ARIMA(1,2,0)预测该地区麻疹的发病趋势,2009～2012年麻疹的发病率分别为2.66/10万、1.65/10万、0.88/10万、0.35/10万。[结论]至2012年该地区达到WHO和我国提出消除麻疹水平(1/100万以下)的要求尚存在一定差距,因此,当前应进一步加强本地区麻疹预防控制措施的力度。     

ARIMA模型在传染病发病率预测中的应用

目的:探讨应用时间序列ARIMA模型进行法定传染病发病率预测的可行性。方法:应用SPSS11.5软件对1986年~2002年逐月发病率进行RIMA模型建模拟合,用所得到的模型对2003年各月发病率进行预测,并与实际发病率进行比较。结果:ARIMA(0,1,1)×(0,1,1)12模型很好地拟合了既往时间段上的发病率序列,对2003年各月发病率的预测值符合实际发病率变动趋势。结论:ARIMA模型能很好地模拟传染病发病率在时间序列上的变动趋势,并对未来的发病率进行预测,为传染病防治工作服务。     

All‐cause mortality and fatal alcohol poisoning in Belarus, 1970–2005

Introduction and Aims . Although alcohol appears to be an important contributor to the burden of disease in the countries of eastern Europe, little systematic research has been undertaken on its impact on mortality in the former Soviet republic of Belarus. There may be a number of factors underlying the particularly negative effect of alcohol on mortality in Belarus, including the pattern of drinking and use of surrogates. A solid body of research and empirical evidence suggests that hazardous patterns of alcohol consumption (binge drinking) lead to quicker and deeper intoxication, increasing the propensity for alcohol‐related mortality. Design and Method . To estimate the aggregate level effect of binge drinking on the all‐cause mortality rate, trends in the all‐cause mortality and fatal alcohol poisoning rates (as a proxy for binge drinking) in Belarus from 1970 to 2005 were analysed employing AutoRegressive Integrated Moving Average (ARIMA) time–series analysis in order to assess a bivariate relationship between the two time–series. Results . The results of time–series analysis suggest a close relationship between all‐cause mortality and fatal alcohol poisoning rates at the population level. Conclusions . This study supports the hypothesis that alcohol and all‐cause mortality are connected closely in countries where the drinking culture is characterised by heavy drinking episodes and adds to the growing body of evidence that a substantial proportion of total mortality in Belarus is due to acute effects of binge drinking.     

传递函数模型在医院业务收入预测中的应用

[目的]介绍传递函数模型在医院业务收入预测中的应用。[方法]利用传递函数模型对医院业务收入影响因素进行分析。在此基础上,利用分析结果对医院业务收入做出预测,并将预测结果与ARIMA模型预测结果进行比较。[结果]在本次研究中,出院人数成为医院业务收入的主要影响因素;预测结果对比中,传递函数模型预测结果较ARIMA模型预测结果接近实际值。[结论]传递函数模型可在对医院业务收入影响因素进行客观分析的基础上作出较为精确的预测。     

An investigation of the effect of privatization of retail sales of alcohol on consumption and traffic accidents in Alberta, Canada

AIMS: Privatization of the retail sale of alcohol in Alberta took place primarily between the end of the 1980s and the beginning of the 1990s. The aim of this study was to evaluate the effects of this privatization on alcohol sales and on the incidence of fatal motor vehicle traffic accidents in the province. DATA AND METHOD: Interrupted time-series analysis (ARIMA) with a quasi-experimental control area design was used, and all series were differenced to remove long-term trends. Canada, with the exception of Alberta, was the control area. The effects of privatization were measured by means of created privatization variables. In the analyses of the effects of privatization on alcohol sales, the inhabitants' disposable income and alcohol prices were used as control variables. The study period was 1950-2000. When effects on the number of fatal motor vehicle traffic accidents were analysed the number of road motor vehicle registrations was used as a control variable, and the study period was 1950-98. FINDINGS: Privatization had a significant permanent effect on the sale of spirits, but the effect was not large enough to affect total sales. The effect on wine and beer sales was not significant. There was no significant effect on the number of fatal motor vehicle traffic accidents. CONCLUSION: The fact that sales on the wholesale level continued to be monopolized, along with the fact that alcohol sales were never allowed in ordinary grocery stores, may explain the lack of any larger effects of privatization on alcohol sales in Alberta.