Skip to main content

Table 2 Perfomance Measures

From: Predicting patient outcomes in psychiatric hospitals with routine data: a machine learning approach

 

Sensitivity / Recall

Specificity

Positive Predictive Value / Precision

Negative Predictive Value

Prevalence

Detection Prevalence

Balanced Accuracy

Precision at least 20%

 Non-Response

1.00

0.00

0.23

1.00

0.23

1.00

0.50

 Coercive Treatment

0.73

0.78

0.20

0.97

0.07

0.26

0.76

 Long LOS

0.98

0.28

0.20

0.99

0.16

0.76

0.63

 Short LOS

0.83

0.37

0.20

0.92

0.16

0.66

0.60

Precision at least 25%

 Non-Response

0.96

0.15

0.25

0.93

0.23

0.87

0.56

 Coercive Treatment

0.48

0.89

0.25

0.96

0.07

0.13

0.69

 Long LOS

0.94

0.48

0.25

0.98

0.16

0.58

0.71

 Short LOS

0.61

0.65

0.25

0.90

0.16

0.39

0.63

Precision at least 33%

 Non-Response

0.52

0.69

0.33

0.83

0.23

0.36

0.61

 Coercive Treatment

0.23

0.97

0.33

0.94

0.07

0.05

0.60

 Long LOS

0.49

0.82

0.33

0.90

0.16

0.23

0.65

 Short LOS

0.41

0.84

0.33

0.88

0.16

0.20

0.62

  1. Outcomes without clinically meaningful operational points are not shown (Crisis Intervention & 1:1 Observation). Actual precision could be more than minimum precision. TP True Positive, FP False Positive, TN True negative, FN False Negative, Sensitivity = TP/(TP+ FN), Specificity = TN/(TN + FP), Positive Predictive Value = TP/(TP + FP), Negative Predictive Value = TN/(TN + FN), Prevalence = (TP + FN)/(TP + FP + TN + FN), Detection Prevalence = (TP + FP)/(TP + FP + TN + FN), Balanced Accuracy = (Sensitivity+Specificity)/2