Table 3 Subgroup analysis results by study characteristics and model types.

From: Global performance of machine learning models to predict all-cause mortality: systematic review and meta-analysis

Variable

AUC (95%CI)

I2 (%)

Sample size

 Less than 2000

0.835 (0.799–0.871)

96.1

 2000 or more

0.830 (0.789–0.870)

100

TRIPOD + AI

 Less than 35 points

0.838 (0.817–0.858)

99.8

 35 or more

0.814 (0.704–0.924)

100

Models

 Tree-based models

0.832 (0.774–0.890)

100

 Neural networks

0.823 (0.793–0.854)

99.7

 Linear/statistical

0.853 (0.759–0.947)

99.2

 Ensemble/hybrid models

0.829 (0.781–0.878)

98

 Other

0.821 (0.734–0.907)

98.3

Imputation of confidence intervals

 Non-imputed

0.815 (0.768–0.862)

100

I mputed

0.856 (0.823–0.890)

99.4

  1. Pooled AUC values with 95% confidence intervals and heterogeneity measures (I2) for subgroup analyses by sample size, TRIPOD + AI quality score, machine learning model categories, and confidence interval imputation status.
  2. Note Other models category includes algorithms used by single studies: DSM (n = 1), DLS-MSM (n = 1), ICISS (n = 1), Support Vector Machine (n = 1), and Bayesian Network (n = 1).
Back to article page