Table 3 Details on the performance evaluation of HUST-19 for the prediction of morbidity outcomes using data from cohort 2

From: Open resource of clinical data from patients with pneumonia for the prediction of COVID-19 outcomes via deep learning

Prediction

Type

AUC

Sn (%)

Sp (%)

Ac (%)

PPV (%)

NPV (%)

MCC

CT based

Control

0.895

53.57%

94.47%

83.70%

77.59%

85.06%

0.5486

Type I

0.775

86.67%

50.36%

70.85%

69.33%

74.47%

0.4027

Type II

0.832

32.73%

95.08%

84.33%

58.06%

87.15%

0.3546

CF based

Control

0.888

54.72%

99.59%

86.04%

98.31%

83.56%

0.6668

Type I

0.834

72.53%

78.70%

75.50%

78.57%

72.68%

0.5124

Type II

0.845

71.63%

78.82%

77.49%

42.45%

92.65%

0.4200

HUST-19

Control

0.944

51.19%

98.30%

85.89%

91.49%

84.93%

0.6150

Type I

0.860

80.56%

76.26%

78.68%

81.46%

75.18%

0.5673

Type II

0.884

80.00%

78.41%

78.68%

43.56%

94.95%

0.4743

  1. The Sn, Sp, Ac, PPV, NPV and MCC values were directly calculated.