Skip to main content

Table 3 Model performance metrics at the first and second stages of stratification for both the 30- and 90-day summative “Any event” outcome

From: Improving event prediction using general practitioner clinical judgement in a digital risk stratification model: a pilot study

Model performance metric

1st stage,

2 groups,

Non-Escalated vs Escalated

2nd stage,

3 groups,

Non-Escalated, GP No Concern, GP Concern

2nd stage,

2 groups,

GP Concern vs No Concern

30-day Any event

 ROC c-statistic

0.610 (0.598—0.632), p < 0.001

0.614 (0.592—0.637), p < 0.001

0.603 (0.565—0.642), p < 0.001

 Sensitivity

34.1%

11.0%

32.3%

 Specificity

87.9%

98.6%

88.4%

 PPV

6.7%

16.7%

16.8%

 NPV

98.1%

97.7%

95.7%

 Accuracy

86.6%

96.4%

84.5%

90-day Any event

 ROC c-statistic

0.586 (0.572—0.600), p < 0.001

0.588 (0.574—0.602), p < 0.001

0.568 (0.542—0.595), p < 0.001

 Sensitivity

28.6%

7.1%

24.7%

 Specificity

88.5%

98.7%

89.0%

 PPV

15.0%

28.3%

28.1%

 NPV

94.6%

93.7%

87.1%

 Accuracy

84.5%

92.7%

79.4%