Table 6.

Experimental results of the imputation performance using CATSI and the two components separately

	PCL	PK	PLCO2	PNA	HCT	HGB	MCV
Recurrent component	0.192	0.238	0.205	0.213	0.212	0.203	0.243
Cross-feature component	0.204	0.267	0.249	0.212	0.110	0.107	0.287
Fusion of both components	0.174	0.243	0.203	0.196	0.144	0.135	0.253
	PLT	WBC	RDW	PBUN	PCRE	PGLU	Mean
Recurrent component	0.179	0.221	0.213	0.152	0.201	0.250	0.209
Cross-feature component	0.295	0.291	0.317	0.271	0.269	0.282	0.243
Fusion of both components	0.187	0.227	0.213	0.157	0.206	0.260	0.200

Experiments are done with the individual missing dataset as in Section 5.1. The numbers in bold indicate the best performance for the corresponding analyte, and those in italics indicate the second best ones. The performance is measured using normalized root mean square deviation (n RMSD)