Table 3.

AUCs for link prediction performance amongst four HPO embeddings with six machine learning algorithms using four edge embeddings operations. (The best AUC for each edge embeddings in bold.)

Operations	Algorithms	HPOEmb-Original	HPOEmb-DECIPHER	HPOEmb-OMIM	HPOEmb-Orphanet
Hadamard	DT	0.69	0.69	0.77	0.8
	LR	0.76	0.76	0.9	0.92
	SVM	0.72	0.73	0.84	0.85
	RF	0.79	0.79	0.9	0.92
	NB	0.73	0.73	0.82	0.85
	MLP	0.72	0.72	0.84	0.86
Average	DT	0.7	0.69	0.76	0.8
	LR	0.79	0.79	0.9	0.92
	SVM	0.75	0.75	0.84	0.86
	RF	0.79	0.8	0.9	0.92
	NB	0.75	0.75	0.84	0.86
	MLP	0.75	0.74	0.84	0.86
L1	DT	0.67	0.66	0.58	0.57
	LR	0.81	0.8	0.71	0.67
	SVM	0.75	0.74	0.66	0.64
	RF	0.8	0.8	0.73	0.71
	NB	0.75	0.75	0.65	0.62
	MLP	0.75	0.75	0.67	0.63
L2	DT	0.67	0.66	0.58	0.57
	LR	0.8	0.8	0.7	0.66
	SVM	0.74	0.74	0.64	0.61
	RF	0.8	0.8	0.73	0.72
	NB	0.74	0.74	0.63	0.6
	MLP	0.75	0.74	0.65	0.63