Table 3 Predictive performance of central venous catheterization according to each modeling method using deep learning or machine learning technique for the combination of various features.

From: Preoperative prediction of the need for arterial and central venous catheterization using machine learning techniques

Features

Model

AUROC

AUPRC

F1-score

Preoperative data

DNN

0.7527 ± 0.0022

0.2004 ± 0.0030

0.0499 ± 0.0091

XGBoost

0.5131 ± 0.0005

0.0807 ± 0.0009

0.0520 ± 0.0017

RF

0.5152 ± 0.0010

0.0792 ± 0.0014

0.0608 ± 0.0037

LR

0.5156 ± 0.0006

0.0820 ± 0.0010

0.0620 ± 0.0023

Laboratory data

DNN

0.6966 ± 0.0054

0.1536 ± 0.0049

0.0026 ± 0.0017

XGBoost

0.5154 ± 0.0006

0.0844 ± 0.0013

0.0608 ± 0.0025

RF

0.5167 ± 0.0008

0.0762 ± 0.0011

0.0681 ± 0.0028

LR

0.5016 ± 0.0001

0.0644 ± 0.0004

0.0080 ± 0.0007

Operation code

DNN

0.9308 ± 0.0012

0.6754 ± 0.0036

0.6400 ± 0.0055

XGBoost

0.6673 ± 0.0016

0.3495 ± 0.0032

0.4918 ± 0.0036

RF

0.5361 ± 0.0118

0.1279 ± 0.0212

0.1241 ± 0.0384

LR

0.6257 ± 0.0014

0.2788 ± 0.0028

0.3962 ± 0.0036

All features

DNN

0.9261 ± 0.0097

0.6849 ± 0.0219

0.3687 ± 0.0658

XGBoost

0.7062 ± 0.0015

0.4146 ± 0.0032

0.5699 ± 0.0032

RF

0.5371 ± 0.0076

0.1283 ± 0.0137

0.1337 ± 0.0255

LR

0.5057 ± 0.0004

0.0683 ± 0.0006

0.0248 ± 0.0016

  1. Data are presented as means (95% confidence intervals).
  2. AUROC, area under receiver operating characteristic; AUPRC, area under precision-recall curve; DNN, deep neural network; XGBoost, extreme gradient boosting; DT, decision tree; RF, random forest; LR, logistic regression; ASA-PS, American Society of Anesthesiologists physical status.
Back to article page