Table 4.

Performance of machine learning models for risk prediction of long-term cancer-specific survival of patients with second primary lung cancer after 2004.

Model			Sensitivity, %		Specificity, %		AUCa (95% CI)
1-year cancer-specific survival
	XGBb	77		60.2		0.73 (0.71-0.75)
	RFCc	76.7		63		0.74 (0.72-0.76)
	ADBd	83.1		54.4		0.75 (0.73-0.77)
	KNNe	70.9		63.6		0.72 (0.70-0.74)
	ANNf	88.2		41.9		0.74 (0.72-0.76)
	GBDTg	90.6		36		0.74 (0.72-0.76)
3-year cancer-specific survival
	XGB	69.9		73.8		0.77 (0.75-0.79)
	RFC	75.6		69.2		0.77 (0.75-0.79)
	ADB	79.3		66.4		0.76 (0.74-0.78)
	KNN	79.6		64		0.75 (0.73-0.77)
	ANN	83.6		59.9		0.77 (0.75-0.79)
	GBDT	84.4		57.6		0.75 (0.73-0.77)
5-year cancer-specific survival
	XGB	79.6		71.3		0.78 (0.75-0.81)
	RFC	79.2		71.5		0.79 (0.76-0.82)
	ADB	75.3		74.7		0.79 (0.76-0.82)
	KNN	74.3		73.9		0.77 (0.74-0.80)
	ANN	79.3		71.5		0.80 (0.77-0.83)
	GBDT	80.1		69.5		0.78 (0.75-0.81)
10-year cancer-specific survival
	XGB	78.8		74.7		0.84 (0.80-0.88)
	RFC	78.3		40.7		0.83 (0.79-0.87)
	ADB	78.4		81		0.84 (0.80-0.88)
	KNN	80.7		73.4		0.78 (0.72-0.84)
	ANN	68.8		88.6		0.85 (0.81-0.89)
	GBDT	79.7		78.5		0.85 (0.81-0.89)

^aAUC: area under the curve.

^bXGB: extreme gradient boosting.

^cRFC: random forest classifier.

^dADB: adaptive boosting.

^eKNN: K nearest neighbor.

^fANN: artificial neural network.

^gGBDT: gradient boosting decision tree.