Preoperative prediction of microsatellite instability status in colorectal cancer based on a multiphasic enhanced CT radiomics nomogram model

Table 4 Detailed performance of each model in the training and testing datasets

Datasets	Models	AUC(95% CI)	Sensitivity	Specificity	Accuracy	PPV	NPV
Training	AP model	0.772(0.701–0.843)	0.720	0.756	0.749	0.434	0.912
	VP model	0.722(0.645–0.799)	0.680	0.705	0.700	0.374	0.895
	DP model	0.750(0.679–0.821)	0.640	0.746	0.724	0.395	0.889
	AP + VP + DP model	0.827(0.763–0.891)	0.760	0.798	0.790	0.494	0.928
	Clinical model	0.765(0.687–0.843)	0.760	0.684	0.700	0.384	0.917
	Nomogram	0.894(0.848–0.939)	0.820	0.819	0.819	0.539	0.946
Testing	AP model	0.760(0.638–0.881)	0.810	0.711	0.731	0.415	0.937
	VP model	0.671(0.552–0.790)	0.952	0.349	0.471	0.270	0.967
	DP model	0.668(0.532–0.805)	0.571	0.783	0.740	0.400	0.878
	AP + VP + DP model	0.714(0.604–0.825)	0.857	0.614	0.663	0.360	0.944
	Clinical model	0.783(0.642–0.923)	0.762	0.819	0.808	0.516	0.932
	Nomogram	0.839(0.738–0.940)	0.868	0.867	0.827	0.560	0.911

AUC, Area under the curve; PPV, Positive predictive value; NPV, Negative predictive value

ISSN: 1471-2342