广义相加模型在细菌性痢疾预警中的应用  

Application of Generalized Additive Model in Early Warning of Bacillary Dysentery

作　　者：陶芳芳[1] 赵耐青[2] 何懿[1] 林庆能[1] 毛智盛[1] 孙晓冬[1] 

机构地区：[1]上海市疾病预防控制中心,200336 [2]复旦大学公共卫生学院,200032

出　　处：《中国卫生统计》2012年第4期481-483,共3页Chinese Journal of Health Statistics

基　　金：上海市公共卫生体系建设新三年行动计划项目

摘　　要：目的探讨应用广义相加模型进行细菌性痢疾预测预警的可行性。方法收集上海市2004～2008年逐日细菌性痢疾发病资料和当地气象数据。采用时间序列的广义相加模型,在控制长期趋势、季节趋势及星期效应等混杂因素的基础上,分析气温、相对湿度、气压等气象因素与细菌性痢疾发病的关系。将广义相加模型得到的预测值的95%范围的上限值作为当前细菌性痢疾发病水平的预警限。结果模型预测的动态趋势与实际情况基本一致,模型的拟合效果较好。2009年细菌性痢疾日报告病例数大多数都落入了预测值的可信区间范围,未达到预警限,与实际流行情况相符。结论广义相加模型可应用于细菌性痢疾预测预警,为疫情防控提供科学依据。Objective To explore application of Generalized additive model in predicting and early warning of incidence of Bacillary Dysentery. Methods Collecting the daily incidence and weather surveil- lance data, the correlation between the weather factors （ temperature, humidi- ty, air pressure） and the daily incidence of residents in Shanghai were ana- lyzed by using the Generalized additive model of time series adjusting for the secular trend, seasonal trend and day of week. Application of the upper limit of the predictive value of 95 % range by generalized additive model will be for the early-warning threshold. Results The trend of model pre- dictions was consistent with actual situation. In 2009 the most of daily inci- dence of Bacillary Dysentery was in a confidence interval range of the pre- dicted value. It did not meet the early-warning threshold and was consistent with the actual prevalence. Conclusion Generalized additive model can be applied in forecasting and early warning of Bacillary Dysentery and it provide the scientific basis for the disease control and prevention.

关 键 词：细菌性痢疾 气象因素 广义相加模型 预警 

分 类 号：R516.4[医药卫生—内科学]