Sayers B M, Segal S D, Henshall W R
Med Inform (Lond). 1978 Mar;3(1):51-68. doi: 10.3109/14639237809016062.
Computer-based statistical signal methods have been applied to the analysis of case-occurrence time and location data from a 12-year complete series of anencephalic stillbirths in the Fylde peninsula of Lancashire. In the absence of any certainty about causative or contributory factors in these occurrences, the primary questions are if the underlying process is Poisson, or if there is evidence of a seasonal influence, and if any regional differences or communicable factors can be identified. With the limited case numbers (124) most small-sample statistical tests were not sufficiently informative. But using pattern and print-process analysis methods and simulation techniques, further elucidation is possible; for example, it is shown that a seasonally rate-modulated Poisson process is a reasonable model for case occurrences in the southern region, but not in the north, for which a different process must apply and where some evidence exists of clustering of preferred inter-event intervals. It is concluded in particular, that there is a regional dependence, with seasonal factors, of the process underlying case occurrences, and in general, that signal analysis methods are useful in the interpretation of small-sample spatio-tempered epidemiological data.
基于计算机的统计信号方法已被应用于分析兰开夏郡菲尔德半岛12年完整系列无脑儿死产病例的发生时间和地点数据。由于这些事件的致病或促成因素尚无定论，主要问题是潜在过程是否为泊松过程，是否有季节性影响的证据，以及是否能识别出任何区域差异或传染性因素。由于病例数量有限（124例），大多数小样本统计检验提供的信息不足。但使用模式和打印过程分析方法以及模拟技术，可以进一步阐明情况；例如，结果表明，季节性速率调制泊松过程是南部地区病例发生的合理模型，但不适用于北部地区，北部地区必须应用不同的过程，并且存在一些事件间偏好间隔聚类的证据。特别得出的结论是，病例发生背后的过程存在区域依赖性，并受季节性因素影响，总体而言，信号分析方法有助于解释小样本时空流行病学数据。
