Skip to main content

Table 1 Study characteristics

From: Using machine-learning risk prediction models to triage the acuity of undifferentiated patients entering the emergency care system: a systematic review

Author Year Country Population Outcome Methods used Predictors Sample size EPV Method of testing
Azeez et al. [25] 2014 Malaysia ED Triage level NN, ANFIS 20 2223   Random split sample (70:30)
Caicedo-Torres et al. [26] 2016 Spain ED Discharge LR, SVM, NN 147 1205   Random split sample (80:20), 10-fCV
Cameron et al. [27] 2015 Scotland ED Hospitalisation LR 9 215231   Random split sample (66:33), bootstrapping (10,000)
Dinh et al. [28] 2016 Australia ED Hospitalisation LR 10 860832 9470 Random split sample (50:50)
Dugas et al. [29] 2016 USA ED Critical illness LR 9 97000000 525 Random split sample (90:10), 10f-CV
Golmohammadi [30] 2016 USA ED Hospitalisation LR, NN 8 7266 460.25 Split sample (70:30)
Goto et al. [31] 2019 USA ED Critical illness, hospitalisation LR, LASSO, RF, GBDT, DNN 5 52037 32.60 Random split sample (70:30)
Hong et al. [32] 2018 USA ED Hospitalisation LR, GBDT, DNN 972 560486 171.44 Random split sample (90:10)
Kim, D et al. [33] 2018 Korea Prehospital Critical illness LR, RF, DNN 5 460865 3583.60 10f-CV
Kim, S et al. [34] 2014 Australia ED Hospitalisation LR 8 100123 1074.86 Apparent performance
Kwon et al. (1) [35] 2018 Korea ED Critical illness, hospitalisation DNN, RF 7 10967518 133667.89 Split sample (50:50), + external validation dataset
Kwon et al. (2) [36] 2019 Korea ED Critical illness, hospitalisation DNN, RF, LR 8 2937078 14047.57 Split sample (50:50)
Levin et al. [37] 2018 USA ED Critical illness, hospitalisation RF 6 172726 56.74 Random split sample (66:33), bootstrapping
Li et al. [38] 2009 USA Pre-hospital Hospitalisation LR, NB, DT, SVM 6 2784   10f-CV
Meisel et al. [39] 2008 USA Pre-hospital Hospitalisation LR 9 401   Bootstrap resampling (1000)
Newgard et al. [40] 2013 USA Prehospital Critical illness CART 40 89261   Cross-validation
Olivia et al. [41] 2018 India ED Triage level DT, SVM, NN, NB 8    10f-CV
Raita et al. [42] 2019 USA ED Critical illness, hospitalisation LR, LASSO, RF, GBDT, DNN 6 135470 107 Random split sample (70:30)
Rendell et al. [43] 2019 Australia ED Hospitalisation B, DT, LR, NN, NB, KNN 11 1721294 5521 10f-CV
Seymour et al. [44] 2010 USA Prehospital Critical illness LR 12 144913 156 Random split sample (60:40)
van Rein et al. [45] 2019 Netherlands Prehospital Critical illness LR 48 6859 3.4375 Separate external validation
Wang et al. [46] 2013 Taiwan ED Triage level SVM 6 3000   10f-CV
Zhang et al. [47] 2017 USA ED Hospitalisation LR, NN 25 47200 91.8 10f-CV
Zlotnik et al. [48] 2016 Spain ED Hospitalisation NN 9 153970 614.5 10f-CV
Zmiri et al. [49] 2012 Israel ED Triage level NB, C4.5 4 402   10f-CV
  1. ANFIS Adaptive Neuro-Fuzzy Inference System, B Bayesian Network, CART Classification and Regression Tree, DT Decision Tree, DNN Deep Neural Network, EPV Events Per Variable, GBDT Gradient Boosted Decision Tree, KNN K-Nearest Neighbours, LR logistic regression, LASSO Least Absolute Shrinkage and Selection Operator, NB Naïve Bayes, NN Neural Network, RF Random Forest, SVM Support Vector Machine