Skip to main content

Table 5 Predictive performance of the simplified model for an extended testing data (COVID-19 confirmed cases during 1 May–9 Aug 2020), which was supplemented to this study when the manuscript was revised in mid-August 2020

From: Development of a data-driven COVID-19 prognostication tool to inform triage and step-down care for hospitalised patients in Hong Kong: a population-based cohort study

 

Simplified model* (based on 7 features)

 

On Day 1 of admission

On Day 5 of admission#

Predicted class

Predicted class

Critical/serious

Stable

Satisfactory

Critical/serious

Stable

Satisfactory

Actual class

Critical/serious

187

13

23

207

7

9

Stable

227

1389

62

187

1480

11

Satisfactory

7

0

1076

18

0

1065

Sensitivity

By outcome class

83.9%

82.8%

99.4%

92.8%

88.2%

98.3%

Macro averaged

88.9%

92.2%

Micro averaged

88.9%

92.2%

Specificity

By outcome class

91.5%

99.0%

95.5%

92.6%

99.5%

98.9%

Macro averaged

97.2%

98.8%

Micro averaged

94.4%

96.1%

Accuracy

88.9%

92.2%

  1. *Model performance based on testing dataset (n = 2984)