Table 3.

Prediction performance for combinations of data sourcing with bag of features+Node2vec and random forest algorithms.

Feature types		AUROCa (%)
			Base feature set		With genetic information
1 feature type
	Gb		73.12		90.89
	Dc		65.01		88.37
	Hd		91.00		95.80
	Le		72.83		89.94
	Mf		73.21		90.92
2 feature types
	DH		91.55		96.09
	DL		77.09		90.88
	DM		91.30		95.92
	HL		71.53		89.02
	MH		91.22		95.75
	ML		91.98		96.01
3 feature types
	DHL		76.76		91.28
	DMH		91.76		96.56
	DML		91.43		95.76
	MHL		91.74		96.19
4 feature types
	DMHL		73.12		90.89

^aAUROC: area under the receiver operating characteristic curve.

^bG: genetic information.

^cD: diagnosis.

^dH: family historical records.

^eL: lab test.

^fM: medication.