Please use this identifier to cite or link to this item:
|Title:||Truncation Methods on the SROC Curve|
|Department:||Mathematics and Statistics|
|Abstract:||The Summary Receiver Operating Characteristic (SROC) curve is a method used to summarize the performance of a diagnostic test using data from a meta-analysis, and the Area Under the SROC Curve (AUC) is a measure of the performance of a diagnostic test. The Partial Area Under the SROC Curve (partial AUC) is sometimes used instead of the AUC, to include only the clinically relevant part. Several truncation methods are used on simulated data to determine the effect of the truncation methods on estimating the properties of partial AUC - mean, bias and standard deviation. Also, when part of the data is truncated before fitting the SROC curve, we examined how the properties of the SROC parameters are affected by the choice of truncation method. The results show that the truncation methods do affect the properties of partial AUC and the estimated SROC parameters. First of all, the estimated values of the partial AUC, with or without scaling, are increased as the value of truncation point increases. The standard deviation of the partial AUC has an increasing relationship with the value of truncation point, and the standard deviation of the scaled partial AUC is in contrast. Truncating a certain percentage of the data performs worse than truncating a part of the SROC curve with respect to the accuracy of estimation. As for estimating the SROC curve parameters a and b, the truncation method which keeps more studies gives more accurate estimation.|
|Appears in Collections:||Open Access Dissertations and Theses|
Files in This Item:
|CAI_JING_201612_MSc.pdf||3.23 MB||Adobe PDF||View/Open|
Items in MacSphere are protected by copyright, with all rights reserved, unless otherwise indicated.