BMC Medical Research Methodology

Table 13 Membership disclosure (Hamming distance=0.1, r=1000) on several synthetic sample sizes

From: Generation and evaluation of synthetic patient data

	# of synthetic samples
	5k		10k		20k		∼170k (all)
	Precision	Recall	Precision	Recall	Precision	Recall	Precision	Recall
IM	0.491	0.842	0.493	0.905	0.493	0.927	0.497	0.97
BN	0.491	0.848	0.496	0.905	0.492	0.93	0.499	0.985
MPoM	0.492	0.872	0.496	0.906	0.498	0.942	0.499	0.99
CLGP	0.496	0.867	0.497	0.913	0.498	0.943	0.500	0.988
MC-MedGAN	0.488	0.585	0.486	0.627	0.482	0.651	0.491	0.751
MICE-LR	0.497	0.868	0.497	0.909	0.500	0.957	0.499	0.988
MICE-LR-DESC	0.489	0.854	0.496	0.909	0.494	0.94	0.500	0.991
MICE-DT	0.496	0.86	0.498	0.924	0.499	0.952	0.502	0.994

BREAST small-set. Values shown are the average performance over 10 independently generated synthetic samples

Back to article page

ISSN: 1471-2288

Contact us

General enquiries: journalsubmissions@springernature.com