Table 2 The efficacy (precision, sensitivity and F1 score) of top three machine-learning classifiers in the training (A) and testing (B) sets (using imputed data).

From: Early predictive values of clinical assessments for ARDS mortality: a machine-learning approach

Datasets

Models

Predictive values for survivors (Day 0 = 498, Day 3 = 516)

Predictive values for non-survivors (Day 0 = 202, Day 3 = 158)

Precision (95% CI)

Sensitivity (95% CI)

F1 score (95% CI)

Precision (95% CI)

Sensitivity (95% CI)

F1 score (95% CI)

A. In the training set

Day 0 (n = 700)

RF

0.85 (0.83, 0.88)

0.98 (0.98, 1.0)

0.91 (0.91, 0.93)

0.96 (0.93, 0.99)

0.59 (0.54, 0.66)

0.73 (0.69, 0.78)

XGBoost

0.87 (0.86, 0.9)

0.98 (0.98, 0.99)

0.92 (0.92, 0.94)

0.95 (0.92, 0.98)

0.66 (0.61, 0.72)

0.78 (0.74, 0.82)

SVM

0.94 (0.93, 0.97)

0.98 (0.98, 1.0)

0.96 (0.96, 0.98)

0.97 (0.95, 0.99)

0.87 (0.83, 0.91)

0.91 (0.89, 0.94)

Day 3 (n = 674)

RF

0.90 (0.88, 0.92)

0.99 (0.98, 1.0)

0.94 (0.93, 0.96)

0.95 (0.92, 0.98)

0.65 (0.6, 0.72)

0.77 (0.73, 0.82)

SVM

0.89 (0.87, 0.92)

0.97 (0.97, 0.99)

0.93 (0.92, 0.95)

0.89 (0.84, 0.94)

0.62 (0.57, 0.69)

0.73 (0.69, 0.78)

SC

0.89 (0.87, 0.92)

0.98 (0.98, 1.0)

0.93 (0.93, 0.95)

0.94 (0.9, 0.98)

0.62 (0.56, 0.68)

(0.7, 0.8)

Datasets

Models

Predictive values for survivors (n, Day 0 = 233, Day 3 = 215)

Predictive values for non-survivors (n, Day 0 = 67, Day 3 = 74)

Precision (95% CI)

Sensitivity (95% CI)

F1 score (95% CI)

Precision (95% CI)

Sensitivity (95% CI)

F1 score (95% CI)

B. In the testing set

Day 0 (n = 300)

RF

0.81 (0.78, 0.85)

0.91 (0.88, 0.94)

0.86 (0.83, 0.89)

0.47 (0.34, 0.61)

0.28 (0.19, 0.38)

0.35 (0.25, 0.45)

LR

0.80 (0.77, 0.85)

0.93 (0.9, 0.95)

0.86 (0.84, 0.89)

0.48 (0.34, 0.63)

0.23 (0.16, 0.33)

0.32 (0.22, 0.42)

MLP

0.84 (0.81, 0.89)

0.8 (0.76, 0.85)

0.82 (0.79, 0.85)

0.41 (0.33, 0.51)

0.49 (0.39, 0.59)

0.45 (0.36, 0.53)

Day 3 (n = 289)

RF

0.82 (0.78, 0.86)

0.95 (0.93, 0.98)

0.88 (0.86, 0.91)

0.76 (0.65, 0.88)

0.39 (0.3, 0.49)

0.51 (0.42, 0.61)

XGBoost

0.82 (0.78, 0.86)

0.93 (0.91, 0.97)

0.87 (0.85, 0.9)

0.69 (0.58, 0.82)

0.40 (0.31, 0.5)

0.51 (0.42, 0.6)

SC

0.82 (0.78, 0.86)

0.95 (0.93, 0.98)

0.88 (0.86, 0.91)

0.76 (0.65, 0.88)

0.40 (0.31, 0.5)

0.53 (0.43, 0.62)

  1. Missing data was imputed using an iterative multivariate imputation technique.
  2. RF random forest, LR logistic regression, SVM support vector machine, MLP multi-layer perceptron, SC stacking classifier.
  3. 95% CI = 95% confidence interval.