Table 3. Heart failure EHRs dataset—comparison of three different graph representation methods.

Metric	Distance threshold	K-nearest neighbors	Similarity
Graph density	$2.29\times {10}^{-2}$	$0.71\times {10}^{-2}$	$1.70\times {10}^{-2}$
Average clustering coefficient	0.34	0.25	0.35
Connected components	49	1	6
Largest component size (%)	87.3%	100.0%	98.8%
Assortativity coefficient	0.31	–0.06	0.60
Community count	50	2	7
Modularity score	0.39	0.39	0.48
Homophily score	0.70	0.80	0.87
Chi-square $p$-value	0.00	0.00	0.000
Homophily Z-score	18.1	14.3	29.75

Note:

Quantitative metrics for each graph construction approach. Each metric is defined in Section Methods.