Multi-omics to predict acute radiation esophagitis in patients with lung cancer treated with intensity-modulated radiation therapy

Table 3 The model performance in the multiple train−test splits model

Model	Metric	Training cohort (\(\mu \pm \sigma\))	95% CI	Testing cohort (\(\mu \pm \sigma\))	95% CI
CFM	AUC	\(0.573\pm 0.026\)	[0.56, 0.58]	\(0.509\pm 0.072\)	[0.48, 0.53]
CFM	ACC	\(0.571\pm 0.041\)	[0.56, 0.58]	\(0.538\pm 0.071\)	[0.51, 0.57]
DFM	AUC	\(0.679\pm 0.027\)	[0.67, 0.69]	\(0.604\pm 0.068\)	[0.58, 0.63]
DFM	ACC	\(0.630\pm 0.049\)	[0.61, 0.65]	\(0.552\pm 0.037\)	[0.54, 0.56]
RFM	AUC	\(0.756\pm 0.023\)	[0.79, 0.80]	\(0.744\pm 0.044\)	[0.73, 0.76]
RFM	ACC	\(0.740\pm 0.024\)	[0.73, 0.75]	\(0.695\pm 0.045\)	[0.68, 0.71]
HFM	AUC	\(0.801\pm 0.022\)	[0.79, 0.81]	\(0.747\pm 0.041\)	[0.73, 0.76]
HFM	ACC	\(0.744\pm 0.030\)	[0.73, 0.75]	\(0.696\pm 0.041\)	[0.68, 0.71]

ACC accuracy, \(\mu\) average, \(\sigma\) standard deviation, CI confidence intervals

ISSN: 2047-783X