摘要: |
目的通过混合样本方法减小检测精度带来的误差,提高对低总体率估计的精度。方法通过公式推导说明对给定的灵敏度和特异度,有其适宜检测的理想总体率值;在给定总体率、灵敏度和特异度下,利用计算机模拟和计算不同混合样本大小下对率估计的平均相对误差。结果当实际总体率小于理想值时,通过样本混合可以调整率值,从而减小检测精度带来的误差。结论在低总体率下,针对给定的灵敏度、特异度,混合样本方法可以极大地提高率的估计精度,且减少检测的次数。 |
关键词: 二项分布 总体率 混合样本 灵敏度 特异度 |
DOI:10.3724/SP.J.1008.2011.072 |
投稿时间:2010-07-30修订日期:2010-12-23 |
基金项目: |
|
Pooled sampling method under given sensitivity and specificity in improving accuracy of point estimator of population rate |
TENG Hai-ying1,ZHANG Luo-man2*,SUN Qing-wen1*,MENG Hong2,SONG Mao-hai3 |
(1. Department of Mathematics and Physics, College of Basic Medical Sciences, Second Military Medical University, Shanghai 200433, China;;2. Department of Health Statistics, Faculty of Medical Service, Second Military Medical University, Shanghai 200433, China;;3. Department of Computer Science, College of Basic Medical Sciences, Second Military Medical University, Shanghai 200433, China) |
Abstract: |
ObjectiveTo use the pooled sampling method to reduce the error caused by the detecting precision, so as to improve the accuracy of population rate estimates. MethodsA formula was deduced on how to obtain the ideal rate under given sensitivity and specificity, and the mean relative errors of rate estimation with different pool sizes were simulated and calculated using software SAS 9.1. ResultsWhen the actual rate was lower than the ideal rate, the errors caused by the detecting precision could be greatly reduced through adjusting the rate of mixed samples. ConclusionWhen the population rate is low, the accuracy of rate estimation can be improved and the numbers of tests can be reduced by using pooled sampling method under given specificity and sensitivity. |
Key words: binomial distribution population rate pooled sample sensitivity specificity |