有序聚类分析及周期图法在猩红热流行周期中的应用研究

Application of Sequential Cluster Analysis and Periodogram Method in Epidemic Trend Analysis of Scarlet Fever

目的探讨合肥市猩红热的流行周期,为猩红热发病预测和早期预警提供理论基础。方法合肥市疾病预防控制中心提供1985-2003年猩红热病例资料的监测数据及2004-2008年网络直报监测数据。使用有序聚类分析对猩红热发病阶段划分类别,使用周期图法提取潜在周期并建立相应的周期函数拟合发病率资料。结果 1985-2008年合肥市共有1996名猩红热病例,年平均发病率为1.9620/10万。24年猩红热发病率波动总体可按有序聚类分为低-高-低-高4个阶段,分别为1985-1988年,1989-1997年,1998-2003年,2004-2008年。其中前3阶段总和为19年,与周期图法检测出猩红热发病率序列存在的第一隐含周期T_1=19相同,此外序列还包含第二隐含周期T_2=5。结论有序聚类分析和周期图法可以运用于猩红热流行周期的识别和提取。

Objective We amid to analyze epidemic trend of scarlet fever in Hefei city and provide predictive methods for early warning of scarlet fever. Methods Surveillance data of scarlet fever from 1985 to 2008 were collected from centers for disease control and prevention of Hefei city. Sequential cluster analysis was used to divide these years into several periods. Perio- dogram method was used to extract potential cycle and fit the time series of scarlet fever. Results There were altogether 1996 cases of scarlet fever in Hefei city from 1985 to 2008. The average incidence of scarlet fever was 1. 9620 per 105. During these years,four clusters were classified by sequential cluster analysis which were 1985 - 1988,1989 - 1997,1998 - 2003,2004 - 2008 ,respectively. The first three clusters were totally 19 years which was equal to the first potential cycle T1 of scarlet fever. In addition,the second potential cycle T2 was equal to 5 in incidence series. Conclusion Sequential cluster analysis and perio- dogram method could be used to extract epidemic cycles of scarlet fever incidence.