BMC Medical Informatics and Decision Making

Table 4 Comparison of model performances incorporating weight transfer(mean±std)

From: Enhancing heart failure treatment decisions: interpretable machine learning models for advanced therapy eligibility prediction using EHR data

	Methods	Accuracy	F1	AUC	AUPRC	Cohen’s D
SVM	standard	0.837 (0.052)	0.657 (0.15)	0.823 (0.091)	0.792 (0.13)
	S-weighted	0.839 (0.038)	0.687 (0.042)	0.896 (0.021)	0.858 (0.052)	1.05
	P-weighted	0.853 (0.051)	0.698 (0.10)	0.882 (0.041)	0.874 (0.046)	0.836
Gaussian NB	standard	0.744 (0.066)	0.603 (0.13)	0.795 (0.10)	0.747 (0.15)
	S-weighted	0.839 (0.034)	0.695 (0.046)	0.695 (0.046)	0.832 (0.083)	1.274
	P-weighted	0.829 (0.025)	0.715 (0.043)	0.871 (0.037)	0.794 (0.077)	1.008
RF	standard	0.844 (0.041)	0.656 (0.14)	0.829 (0.096)	0.803 (0.13)
	S-weighted	0.839 (0.034)	0.686 (0.046)	0.889 (0.032)	0.850 (0.054)	0.839
	P-weighted	0.852 (0.051)	0.701 (0.10)	0.888 (0.030)	0.867 (0.042)	0.830

*S-weighted The weights learned from Standard LTR were used to generate visit representation, P-weighted The weights learned from Personalized LTR were used to generate visit representation; Cohen’s D is calculated by comparing the AUC values of the standard model to its weight transferred models

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com