. 2022 Jun 22;12:10537. doi: 10.1038/s41598-022-14422-4

Table 2.

Results of prediction need of CrIs AUROC and AUPRC with 95% CIs according to the machine learning method (XGBoost).

	AUROC (CI)	AUPRC (CI)	Sen (95% CI)	Spec (95% CI)	PPV (95% CI)	NPV (95% CI)
A-line	0.913 (0.899–0.927)	0.121 (0.112–0.130)	0.866 (0.822–0.907)	0.821 (0.785–0.854)	0.042 (0.035–0.049)	0.998 (0.998–0.999)
Oxygen therapy	0.909 (0.904–0.916)	0.576 (0.570–0.583)	0.812 (0.780–0.846)	0.853 (0.819–0.881)	0.313 (0.275–0.353)	0.982 (0.979–0.985)
HFNC	0.962 (0.948–0.976)	0.207 (0.189–0.230)	0.922 (0.873–0.964)	0.906 (0.865–0.941)	0.043 (0.029–0.061)	0.999 (0.999–0.999)
Intubation	0.945 (0.932–0.958)	0.193 (0.180–0.203)	0.891 (0.817–0.940)	0.865 (0.818–0.946)	0.047 (0.032–0.091)	0.999 (0.998–0.999)
MTP	0.920 (0.849–0.991)	0.014 (0.011–0.018)	0.878 (0.722–1.00)	0.896 (0.871–0.982)	0.005 (0.002–0.015)	0.999 (0.999–1.00)
Inotropics and vasopressors	0.899 (0.888–0.911)	0.388 (0.379–0.399)	0.826 (0.783–0.863)	0.827 (0.788–0.868)	0.104 (0.08–0.125)	0.995 (0.993–0.996)

AUROC (CI)

AUPRC (CI)

Sen (95% CI)

Spec (95% CI)

PPV (95% CI)

NPV (95% CI)

A-line

0.913

(0.899–0.927)

0.121

(0.112–0.130)

0.866

(0.822–0.907)

0.821

(0.785–0.854)

0.042

(0.035–0.049)

0.998

(0.998–0.999)

Oxygen

therapy

0.909

(0.904–0.916)

0.576

(0.570–0.583)

0.812

(0.780–0.846)

0.853

(0.819–0.881)

0.313

(0.275–0.353)

0.982

(0.979–0.985)

HFNC

0.962

(0.948–0.976)

0.207

(0.189–0.230)

0.922

(0.873–0.964)

0.906

(0.865–0.941)

0.043

(0.029–0.061)

0.999

(0.999–0.999)

Intubation

0.945

(0.932–0.958)

0.193

(0.180–0.203)

0.891

(0.817–0.940)

0.865

(0.818–0.946)

0.047

(0.032–0.091)

0.999

(0.998–0.999)

MTP

0.920

(0.849–0.991)

0.014

(0.011–0.018)

0.878

(0.722–1.00)

0.896

(0.871–0.982)

0.005

(0.002–0.015)

0.999

(0.999–1.00)

Inotropics and vasopressors

0.899

(0.888–0.911)

0.388

(0.379–0.399)

0.826

(0.783–0.863)

0.827

(0.788–0.868)

0.104

(0.08–0.125)

0.995

(0.993–0.996)

Cut-off value of prediction model for calculating Spec, Sen, PPV, NPV was set by youden index. XGBoost extreme gradient boosting, HFNC high-flow nasal cannula, MTP massive transfusion, AUROC area under the receiver operating characteristic curve, AUPRC area under the precision-recall curve, CI confidence interval, Sen Sensitivity, Spec Specificity, PPV Positive predict value, NPV Negative Predict value.