论文部分内容阅读
目的探索将前瞻性时空扫描统计量运用于传染病预警中适宜基线的选择方法。方法根据当年流行情况选择发病高峰时段结合固定长度基线以及可变长度基线运用前瞻性时空扫描统计量进行模拟实时预警分析,通过不同长度基线的结果比较,探索不同基线在不同流行状况的适用差异并寻找选择适当基线的方法。结果固定长度基线产生了18个信号,而采用可变长度基线只纳入高峰期数据的预警则产生了28个信号,且这28个信号中完全包括了固定长度基线产生的18个信号。其中2009年4月17日及18日固定长度基线的1号信号均被以可变长度基线的时空扫描统计量分为3个较小但发病率更高的信号。结论在发病高峰期选择可变长度基线能够更加精确地区分聚集性区域,并且在经过排除重叠信号后,可以得到更高的探测效能。
Objective To explore a suitable selection method of using forward-looking spatio-temporal scanning statistics in early warning of infectious diseases. Methods According to the prevalence of the year, we selected the peak incidence period combined with the fixed length baseline and the variable length baseline using the prospective space-time scanning statistics to simulate the real-time early warning analysis. By comparing the results of different length baselines, we explored the applicable differences of different baseline in different epidemic situations Find ways to choose the right baseline. Results Fixed length baseline produced 18 signals, whereas early warning with variable length baseline only included peak data resulted in 28 signals that fully contained 18 signals from a fixed length baseline. One of the fixed-length Baseline 1 signals on April 17 and 18, 2009 was divided into three smaller but higher-incidence signals using variable-length baseline spatio-temporal scan statistics. Conclusions Selecting variable length baselines at the peak of the episode allows for a more accurate discrimination of aggregated regions and results in higher detection performance after exclusion of overlapping signals.