Development and validation of an early diagnosis model for severe mycoplasma pneumonia in children based on interpretable machine learning

Table 2 Variable screening in the lasso regression analysis of variables to distinguish SMPP

Variable	Min	lse
Gender	0.064
age	−0.203
S100 A8/A9	5.821	2.801
Tracheal computed tomography	1.273	1.03
Pulmonary radiograph finding	0.289	0.154
Fever duration before admission	0.059	.
Cough duration before admission	0.054	.
Peak body temperature	0.337	0.079
PLT	.	.
PCTper	0.236
MPV	.	.
PLCR	0.146	0.082
TP	−0.018	.
ALB	.	.
ALP	.	.
PA	0.003	.
UA	−0.002	.
CysC	−0.027	.
RBP	−0.26	−0.053
Na⁺	.	.
Ca²⁺	−0.018	.
PT-INR	.	.
D-dimer	0.049	.
AT	.	.
APTT	0.057	0.007
TT	.	.
PCT	.	.
ferritin	0.002	.
hs-CRP	0.003	0.002
IL-6	.	.
IL-10	.	.
IFN-r	.	.
CD³⁺CD⁴⁺T cell	.	.
CD³⁺CD⁸⁺T cell	0	.
CD¹⁶⁺CD⁵⁶⁺ cell	.	.
CD⁴⁺CD²⁵⁺Treg cell	−0.002	.
NLR	−0.073	.

Value in each column is the coefficient value of variable under the two modes of Lasso regression (min = lambda min, 1 se = lambda 1 se). Symbol “.” represent 0 which means such variable was unimportant thus its coefficients was compressed to zero by lasso regression

ISSN: 1465-993X