3.3. The SROC and AUC of Pooled Studies
In 1993, Moses et al. [32] developed the summary receiver operating characteristic (SROC) curve,
which is used to summarize the results from several independent studies for the same biomarker or the
same test. In fact, the ROC curve represents a diagnostic test’s sensitivity versus its false positive rate
(1-specificity). Although SROC and ROC curves are both plotted with sensitivity and 1-specificity, they
are very different. The points of a ROC curve are usually obtained from a single study by changing the
cutoff points continually, while the points of a SROC curve are from independent studies, and each
point represents one study. After two decades of development, while there are more complex models
for obtaining SROC curves to summarize independent studies, most curves are similar to the curve
from Moses’ model [43]. Therefore, to generate a SROC curve, Moses’ model is still the most popular
model. Following the process of Moses’ model, in the first fitting process, two points were outliers.
To keep HCC data, we chose HBsAg negative people (FP = 1, TN = 5) as controls from all healthy
controls in this study [33]. In the second fitting process, only one data point was an outlier.
3.3. The SROC and AUC of Pooled StudiesIn 1993, Moses et al. [32] developed the summary receiver operating characteristic (SROC) curve,which is used to summarize the results from several independent studies for the same biomarker or thesame test. In fact, the ROC curve represents a diagnostic test’s sensitivity versus its false positive rate(1-specificity). Although SROC and ROC curves are both plotted with sensitivity and 1-specificity, theyare very different. The points of a ROC curve are usually obtained from a single study by changing thecutoff points continually, while the points of a SROC curve are from independent studies, and eachpoint represents one study. After two decades of development, while there are more complex modelsfor obtaining SROC curves to summarize independent studies, most curves are similar to the curvefrom Moses’ model [43]. Therefore, to generate a SROC curve, Moses’ model is still the most popularmodel. Following the process of Moses’ model, in the first fitting process, two points were outliers.To keep HCC data, we chose HBsAg negative people (FP = 1, TN = 5) as controls from all healthycontrols in this study [33]. In the second fitting process, only one data point was an outlier.
การแปล กรุณารอสักครู่..
