Evaluation of the models generated from clinical features and deep learning-based segmentations: Can thoracic CT on admission help us to predict hospitalized COVID-19 patients who will require intensive care?

Table 1 Demographics, symptoms, findings, comparison between the groups and discrimination assessment

Parameter	Group 1	Group 2	p	Cut-off or Risk factor	Sensitivity %	Specificity %	AUC
Age (years)	62 (21)	70 (23.3)	< 0.001^a	> 71	55.9	73.1	0.679^d
Sex	Male: 52.0%	Male: 48.0%	0.005^b	Male	67.3	50.3	0.774^e
Sex	Female: 69.3%	Female: 30.7%	0.005^b	Male	67.3	50.3	0.774^e
Fever (body temperature > 38 °C)	58.0%	72.0%	0.014^b	Present	72.0	42.0	0.770^e
Dyspnea	31.9%	43.8%	0.082^b
Cough	49.5%	50.5%	0.491^b
Diabetes mellitus	28.7%	34.6%	0.344^b
Hypertension	42.5%	57.5%	0.210^b
Acute renal failure	1.9%	3.7%	0.446^c
Chronic renal failure	0.4%	3.8%	0.001^c	Present	9.4	99.3	0.863^e
Coronary heart disease	21.0%	33.6%	0.022^b	Present	33.6	79.0	0.764^e
COPD	15.3%	15.0%	0.941^b
Malignity	3.2%	6.5%	0.199^b
CS treatment at the ward	R: 66.7%	R: 83.8%	0.009^b	Need for CS treatment	83.78	33.33	0.595^e
CS treatment at the ward	N/R: 33.3%	N/R: 16.2%	0.009^b	Need for CS treatment	83.78	33.33	0.595^e
High dose CS pulse treatment at the ward	R: 30.9%	R: 63.8%	< 0.001^b	Need for pulse CS treatment	63.77	69.07	0.680^e
High dose CS pulse treatment at the ward	N/R: 69.1%	N/R:36.2%	< 0.001^b	Need for pulse CS treatment	63.77	69.07	0.680^e

Group results were given as Median (IQR)
Group 1: Patients discharged from inpatient floor; Group 2: Patients transferred to ICU
COPD chronic obstructive pulmonary disease, CS corticosteroid, R received, N/R not received
^aMann–Whitney test result
^bChi-squared test result
^cFisher’s exact test result
^dROC analysis results
^eOne-parameter logistic regression result

ISSN: 1471-2342