论文部分内容阅读
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。
Objective To reduce the error caused by the detection accuracy by the method of mixed samples and improve the accuracy of the estimation of low population rate. The method derives from the formula that there is an ideal overall rate of fitness for a given sensitivity and specificity. Using a computer simulation and calculation of the rate estimates under different mixed sample sizes for a given overall rate, sensitivity and specificity Average relative error. Results When the actual overall rate is less than the ideal value, the sample rate can be adjusted by sample mixing to reduce the error caused by the detection accuracy. Conclusions At a low overall rate, for a given sensitivity, specificity, mixed sample approach can greatly improve the accuracy of rate estimation and reduce the number of detections.