TABLE 2.

Number of features by domain, subdomain, and feature origin (claims data or biomedical background knowledge).

Domain	Subdomain	Feature origin		#Time-dependent features (per 30-day time interval)	#static features
		Claims data	Background knowledge
Diagnosis	Biomarker		X	699
Diagnosis	GO term		X	3,759
Diagnosis	MeSH disease		X	137
Diagnosis	Pathway		X	453
Diagnosis	PheWAS	x		1,196
Diagnosis	Symptoms		X	235
General	Age, sex, insurance, hospital	x			17
General	Medical risk		X		53
General	Region	x			403
Drug	GO term		X	1,240
Drug	Pathway		X	435
Drug	PheWAS (known indications)		X	986
Drug	Side effects x frequency		x	1,489
Drug	Substance name	x		453
Drug	Target tissue		x	32
Drug	Therapeutic class		x	104
Drug	Therapeutic group		x	19
				$\mathrm{\sum }=\mathrm{11,237}$	$\mathrm{\sum }=437$

Bold values indicate that the “Feature origin” is “Claims Data”.