Table 3.

CHARMS analysis - Part II (Datasets 6–9).

Domain	Items $\setminus$ Datasets:	BIMCV-COVID19 a			COVIDGR b		RICORD c		AIforCOVID d
Participant data	Eligibility recruitment	Yes			Unclear		Yes		Yes
	Participant description	Age and sex			No		Age, sex, testing method		Sex, age, symptoms, comorbidities
	Treatment info	No			No		No		Yes
	Study dates	Yes			No		No		No
	Country	Spain			Spain		Turkey, USA, Canada, Brazil		Italy
Outcome information	Candidate outcomes	radiological diagnosis	diagnostic test	CARINGe radiological diagnosis	diagostic test	radiological severity	radiological diagnosis	radiological severity	clinical outcome
	Labels	radiological findings (336)	PCR, IgG, IgM f	radiogical findings (22) + BBoxes	PCR	normal, mild, moderate, severe	typical, uncertain atypical, negative	mild, moderate, severe	mild, severe, death
	Labelled images	23k	–	1749	–	426	998	998	–
	Method definition	Yes	Yes	Yes	Yes		Yes	Yes	Yes
	Outcome timing	–	–	–	No		-	Yes	No
	Diagnosis/Prognosis	Both	Diag.	Both	Diag.		Both	Both	Prog.
Predictor information	Candidate predictors	DICOM header					DICOM header		DICOM header, comorbidities, medications, laboratory data, vital signs.
	Method definition	–			-		-		Yes
	Measurement timing	Yes			-		Yes		No
Sample size	# subjects / images	4706/16840(COVID+) 3238/6540 (COVID-)			426/426 (COVID+) 426/426 (Control -)		361/998		983/983

^aBIMCV-COVID19, Valencian Region Medical ImageBank (Vayá et al., 2020)) (https://bimcv.cipf.es/bimcv-projects/bimcv-covid19/).

^bCOVIDGR (Tabik et al., 2020)(https://dasci.es/transferencia/open-data/covidgr-2/).

^cRICORD (Tsai et al., 2021)(https://wiki.cancerimagingarchive.net/pages/viewpage.action?pageId=80969742).

^dAIforCOVID (Soda et al., 2020),(https://aiforcovid.radiomica.it).

^eCARING, external annotation (Mittal et al., 2021) (https://osf.io/b35xu/).

^fImmunoglobulin G (IgG) and Immunoglobulin M(IgM).