Skip to main content
. 2020 Jun 18;26:e926096-1–e926096-11. doi: 10.12659/MSM.926096

Figure 3.

Figure 3

Performance of the EDLC-TN in identification of thyroid cancer in different datasets. (A) Performance of the EDLC-TN on the training dataset. The accuracy, sensitivity and specificity were 93.70%, 93.19%, and 94.01%, respectively. (B) Diagnostic performance of the EDLC-TN and four other state-of-the-art machine learning algorithms. The EDLC-TN demonstrated the highest value for AUC (0.941, 95% CI: 0.935–0.946), sensitivity (93.77%), specificity (94.44%), and accuracy (98.51%). (C) The performance of EDLC-TN on the external validation dataset. The EDLC-TN achieved an accuracy of 95.76%, with a sensitivity of 95.88%, a specificity of 93.75% and an AUC of 0.979 (95% CI: 0.958–0.992).