Table 4.

The performance of the radiomics prediction model for predicting differentiation degree (DD) and lymph node metastases (LNM) of extrahepatic cholangiocarcinoma (ECC) by using a particle swarm optimization and support vector machine (PSO-SVM) model.

Evaluation indicators (%)	DD of ECC		LNM of ECC
	Training group	Testing group	Training group	Testing group
Average AUCa	89.1b	84.6	90.4b	88.9
Average accuracy	82.6	80.9	83.6	81.2
Average sensitivityc	80.5	78.1	85.8	83.2
Average specificityd	83.1	81.5	82.1	79.6
Average PPVe	77.2	75.6	79.1	76.9
Average NPVf	84.6	81.8	89.5	86.8

^aAUC: area under the curve.

^bP<.001.

^cSensitivity is computed at average radiologist specificity.

^dSpecificity is computed at average radiologist sensitivity.

^ePPV: positive predictive value; positive predictive value is computed at average radiologist sensitivity.

^fNPV: negative predictive value.