. Author manuscript; available in PMC: 2018 Apr 24.

Published in final edited form as: Proc SPIE Int Soc Opt Eng. 2018 Mar 7;10577:105770J. doi: 10.1117/12.2293954

Table 3.

Predictive performance of two models and feature selection results

	Feature number	Ridge-LR	Lasso-LR	p (Ridge-LR vs Lasso-LR)	Number of features selected using Lasso-LR
diagnoses	9,146	0.733	0.761	0.0162	115
symptoms	7,738	0.779	0.808	0.169	5
procedures	5,850	0.763	0.785	0.0248	36
medications	2,838	0.745	0.744	0.183	18
laboratory results	3,218	0.713	0.754	<0.001	63
entire EHR features	28,890	0.818	0.839	0.203	41