Table 3.

Prevalence and prediction performance (AUROC and AUPRC) on test sets of Stanford cohorts and MIMIC-III dataset based on TF-IDF, W2V, and D2V

Method	Code	Cohort 1			Cohort 2			MIMIC-III
		Prevalence	AUROC	AUPRC	Prevalence	AUROC	AUPRC	Prevalence	AUROC	AUPRC
TF-IDF	I21	0.25%	0.8977	0.0865	0.28%	0.9499	0.2956	10.36%	0.8648	0.6229
	I25	5.62%	0.9509	0.6797	4.55%	0.9690	0.7369	26.64%	0.9027	0.7945
	I27	1.26%	0.9560	0.5391	0.82%	0.9698	0.5432	4.95%	0.7952	0.2666
	I42	2.20%	0.9790	0.6163	2.01%	0.9810	0.6305	3.92%	0.8507	0.4326
	I48	5.79%	0.9759	0.7941	5.37%	0.9793	0.8072	32.16%	0.8563	0.8084
	I50	2.96%	0.9567	0.5082	2.75%	0.9732	0.6195	25.98%	0.8684	0.7479
	I70	0.15%	0.9282	0.2014	0.24%	0.9520	0.3144	3.61%	0.7982	0.2353
	I85	0.02%	0.9975	0.1315	0.10%	0.9915	0.3759	1.77%	0.9790	0.7298
W2V	I21	0.25%	0.8731	0.0273	0.28%	0.9170	0.0965	10.36%	0.8630	0.5681
	I25	5.62%	0.9315	0.5602	4.55%	0.9405	0.5535	26.64%	0.8659	0.7548
	I27	1.26%	0.9268	0.3785	0.82%	0.9407	0.2858	4.95%	0.7361	0.1563
	I42	2.20%	0.9418	0.3836	2.01%	0.9467	0.3718	3.92%	0.7335	0.1193
	I48	5.79%	0.9493	0.6278	5.37%	0.9524	0.6271	32.16%	0.8103	0.7158
	I50	2.96%	0.9214	0.3355	2.75%	0.9441	0.3871	25.98%	0.8075	0.6200
	I70	0.15%	0.8804	0.0512	0.24%	0.9306	0.1189	3.61%	0.7512	0.1331
	I85	0.02%	0.9969	0.0628	0.10%	0.9805	0.1823	1.77%	0.9545	0.5180
D2V	I21	0.25%	0.8040	0.0167	0.28%	0.8980	0.0613	10.36%	0.8171	0.4533
	I25	5.62%	0.9209	0.5301	4.55%	0.9163	0.4415	26.64%	0.8494	0.6795
	I27	1.26%	0.9033	0.3423	0.82%	0.9307	0.3149	4.95%	0.6911	0.1001
	I42	2.20%	0.9468	0.4451	2.01%	0.9530	0.4298	3.92%	0.7731	0.1699
	I48	5.79%	0.9395	0.5566	5.37%	0.9455	0.5943	32.16%	0.7914	0.6378
	I50	2.96%	0.9185	0.2700	2.75%	0.9271	0.3067	25.98%	0.7660	0.5125
	I70	0.15%	0.8079	0.0150	0.24%	0.9238	0.0872	3.61%	0.7637	0.1561
	I85	0.02%	0.9673	0.0033	0.10%	0.9822	0.2102	1.77%	0.9591	0.4543