Skip to main content
. 2021 Mar 24;13(7):1493. doi: 10.3390/cancers13071493

Figure 3.

Figure 3

Training and validation of the circulating miRNA-based diagnostic classifier. (A) Cancer prediction models including one to six miRNAs (selected by the recursive elimination method) previously detected at higher levels in the blood samples of CRC patients. Representative graphs of overall yield accuracy and LOOCV estimative. (B) Application of the four-miR-based classifier (miR-106a + let-7e + miR-28 + miR-542) in the screening phase (evaluated by the TLDA assay). (C) The four-miR-based classifier applied to a subset of cases of the discovery set (screening phase) using individual RT-qPCR assays. (D) Application of the four-miR-based classifier to a group of samples independent of the screening phase (validation set) using individual RT-qPCR assays. The dotted line indicates the threshold above which a malignant status would be predicted. SVM: support vector machine; LOOCV: leave-one-out cross-validation; AUC: area under the ROC curve; CI95%: 95% confidence interval.