Skip to main content

Table 2 Testing results of the three knowledge incorporation models in comparison with other risk scores and machine learning methods

From: A pattern-discovery-based outcome predictive tool integrated with clinical data repository: design and a case study on contrast related acute kidney injury

Model

AUC

Sensitivity

Specificity

F-score

G-mean

(1) Pre-mode

0.77

0.83

0.57

0.17

0.69

(2) In-mode

0.80

0.70

0.80

0.26

0.75

(3) Post-mode

0.82

0.60

0.88

0.32

0.73

Mehran’s (> 7.8)

0.70

0.24

0.94

0.20

0.47

Chen's (≥ 13)

0.72

0.42

0.88

0.24

0.61

Gao's (> 5)

0.67

0.34

0.94

0.29

0.57

AGEF (≥ 0.66)

0.62

0.37

0.88

0.21

0.57

Logistic regression

0.59

0.84

0.33

0.12

0.53

Decision tree

0.58

0.61

0.55

0.12

0.58

Random forest

0.64

0.58

0.72

0.17

0.64

Easy ensemble

0.70

0.61

0.79

0.23

0.69

  1. The evaluation metrics are defined as follows:
  2. Specificity = TN/(TN + FP); Sensitivity = TP/(TP + FN); Precision = TP/(TP + FP); F-score = 2*Precision*Recall/(Precision + Recall) if TP > 0 and 0 if TP = 0; TP is the count of true positives, FP of false positive, TN of true negatives and FN of false negatives
  3. AUC, areas-under-curve; AGEF, Age, Glomerular filtration rate and Ejection Fraction