Skip to main content

Table 4 Number and proportion of papers according to outcome predicted and measure of predictive accuracy reported (for studies that validated predictions)

From: Use of machine learning to analyse routinely collected intensive care unit data: a systematic review

 

Measure of predictive accuracy reporteda

Outcome predicted

Total papers

AUC and accuracy/sensitivity/specificity

AUC only

Accuracy/sensitivity/specificity only

R 2

Otherb

Complication

73 (45.3%)

24 (32.9%)

17 (23.3%)

28 (38.4%)

 

4 (5.5%)

Mortality

68 (42.2%)

16 (23.5%)

31 (45.6%)

18 (26.5%)

 

3 (4.4%)

Length of stay

18 (11.1%)

2 (11.8%)

3 (16.7%)

5 (27.8%)

8 (44.4%)

1 (5.6%)

Health improvement

16 (10%)

1 (6.3%)

3 (18.8%)

11 (68.8%)

 

1 (6.3%)

Total

161

43 (26.7%)

54 (33.5%)

62 (38.5%)

8 (5.0%)

9 (5.6%)

  1. aPapers can have more than one approach, so percentages may total more than 100. The total of these columns does not account for duplicates as papers can fluctuate how they discuss different results
  2. b“Other” measures of predictive accuracy (number): congruence of ML and clinician’s decisions (1), Matthews correlation coefficient (1), mean absolute differences between observed and predicted (1), mean error rate (1), MSE as loss function (1), Pearson correlation between estimate and actual (1), ratio of wins vs loses against logistic regression (1), rules developed by ML (1)