Table 3.

The predictive performance of individual feature on clinical, semantic, and radiomics model on the training dataset

Individual features	ACC (95% CI)	Sensitivity	Specificity	AUC (95% CI)
Clinical
Age	0.57 (0.51–0.62)	0.28	0.76	0.63 (0.57–0.69)
Gender	0.60 (0.55–0.66)	0	1	0.56 (0.51–0.61)
Smoking	0.60 (0.55–0.66)	0	1	0.55 (0.51–0.58)
Family history	0.60 (0.55–0.66)	0	1	0.50 (0.49–0.52)
Semantic
Diameter	0.71 (0.66–0.76)	0.60	0.78	0.81 (0.77–0.86)
Location	0.57 (0.52–0.63)	0.02	0.94	0.51 (0.44–0.57)
Nodule type	0.77 (0.72–0.82)	0.69	0.83	0.80 (0.76–0.84)
Radiomics
LocInt_peakLocal	0.68 (0.64–0.73)	0.60	0.74	0.83 (0.80–0.87)
Wavelet_HLL_Stats_max	0.68 (0.63–0.72)	0.59	0.73	0.85 (0.82–0.89)
GLRLM_LGRE	0.76 (0.72–0.80)	0.68	0.81	0.90 (0.87–0.92)
Wavelet_LLL_Stats_cov	0.73 (0.68–0.77)	0.63	0.79	0.87 (0.84–0.90)

ACC, accuracy; AUC, area under curve; CI, confidence interval