Skip to main content

Table 22 Performance of 30-day mortality prediction by multiple classifiers

From: An evaluation of time series summary statistics as features for clinical prediction tasks

Combination

AUROC

AUPRC

AUROC

AUPRC

 

Logistic regression

Random forest

mean

0.7257 ±0.0024

0.5118 ±0.0005

0.7716 ±0.0033

0.5314 ±0.0049

first

0.7321 ±0.0029

0.5139 ±0.0027

0.7597 ±0.0036

0.5281 ±0.0069

min, max

0.7376 ±0.0022

0.5140 ±0.0018

0.7760 ±0.0021

0.5351 ±0.0031

min, max, mean

0.7380 ±0.0035

0.5148 ±0.0038

0.7734 ±0.0017

0.5353 ±0.0041

min, max, mean, std

0.7404 ±0.0017

0.5184 ±0.0034

0.7770 ±0.0016

0.5430 ±0.0058

min, max, mean, CV, skew, first

0.7403 ±0.0041

0.5178 ±0.0045

0.7780 ±0.0028

0.5351 ±0.0061

 

SVM

Decision tree

mean

0.7322 ±0.0015

0.5162 ±0.0013

0.5778 ±0.0029

0.3550 ±0.0010

first

0.7164 ±0.0012

0.5055 ±0.0046

0.5102 ±0.0035

0.3069 ±0.0018

min, max

0.7282 ±0.0011

0.5034 ±0.0041

0.5566 ±0.0010

0.3444 ±0.0048

min, max, mean

0.7395 ±0.0033

0.5121 ±0.0046

0.5756 ±0.0034

0.3536 ±0.0012

min, max, mean, std

0.7218 ±0.0024

0.5039 ±0.0013

0.5867 ±0.0003

0.3619 ±0.0005

min, max, mean, CV, skew, first

0.7454 ±0.0015

0.5271 ±0.0035

0.5831 ±0.0026

0.3592 ±0.0024