. 2021 Jan 1;18(1):120–127. doi: 10.7150/ijms.47193

Table 4.

Risk factors associated with the severity of patients with COVID-19

Demographics and clinical characteristics	Univariable OR (95% CI)	P value	Regression coefficient	Feature weight
Age, years*	1.086 (1.050-1.124)	<0.001	0.08
Chest distress	5.133 (2.592-10.168)	<0.001	1.05^a	0.41
Fatigue	3.109 (1.557-6.210)	0.001	0.55
Anorexia	3.501 (1.816-6.749)	<0.001	0.20
Laboratory findings
WBC (G/L)*	1.27 (1.08-1.49)	0.003
N percentage (%)*	1.10 (1.07-1.14)	<0.001	0.19
N (G/L)*	1.27 (1.09-1.48)	0.002	0.29
L percentage (%)*	0.88 (0.84-0.92)	<0.001	0.72
L (G/L)*	0.10 (0.05-0.24)	<0.001	0.46
TB (µmol/L)*	1.125 (1.047-1.209)	0.001
TNI (ng/L)*	1.113 (1.048-1.181)	<0.001	3.87^a	0.74
CysC (mg/L)*	3.079 (1.441-6.579)	0.004	0.33
A/G*	0.031 (0.008-0.114)	<0.001	0.43
ALB (g/L)*	0.800 (0.738-0.867)	<0.001	0.32
Ca (mmol/L)*	0.001 (0.000-0.012)	<0.001	0.82^a	-0.38
CK (U/L)*	1.007 (1.003-1.012)	0.002	0.45
LDH (U/L)*	1.012 (1.007-1.017)	<0.001
ESR (mm/h)*	1.024 (1.007-1.041)	0.005
PCT (µg/L)*	1.2e4 (22-7.4e6)	0.004	1.19^a	0.96
CRP (mg/L)*	1.061 (1.038-1.085)	<0.001	1.17^a	1.13
IL-6 (pg/ml)*	1.115 (1.074-1.157)	<0.001	11.97^a	2.13
IL-10 (pg/ml)*	1.474 (1.166-1862)	0.001

Note. *Per 1 for unit increase. The regression coefficient indicated the weights of the selected features in the multivariate logistic regression with L1 regularization and feature standardization. The parameter C had a value of 3.999 in this regression with L1 regularization.

a: The features that were chosen for a prognostic prediction of severity for patients with moderate COVID-19. The feature weight indicated the weight for each feature used in the probability calculation formula for the prediction of severity in the prognostic model of a linear kernel SVM classifier.