Table 3 Sensitivity and specificity analysis of the reference and machine learning models in the overall and selected validation set

In-hospital mortality

    ANN classifier

93.40%

94.60%

93.30%

92.46%

85.43%

92.84%

(93.1% to 93.8%)

(93.1% to 95.9%)

(93.02% to 93.7%)

(92.08% to 92.83%)

(83.09% to 87.56%)

(92.46% to 93.21%)

    RF classifier

93.02%

87.90%

93.30%

92.52%

87.53%

92.78%

(92.6% to 93.3%)

(85.9% to 89.8%)

(92.9% to 93.6%)

(92.14% to 92.89%)

(85.27% to 89.55%)

(92.40% to 93.15%)

    LR

92.50%

79.70%

93.40%

92.36%

84.93%

92.76%

(92.1% to 92.9%)

(77.4% to 81.9%)

(93.09% to 93.8%)

(91.98% to 92.73%)

(82.55% to 87.11%)

(92.38% to 93.13%)

Cardiac arrest

    ANN classifier

97.10%

93.40%

97.20%

95.93%

86.40%

96.14%

(96.8% to 97.3%)

(91.15% to 95.2%)

(97.0% to 97.4%)

(95.64% to 96.20%)

(82.74% to 89.53%)

(95.86% to 96.41%)

    RF classifier

96.80%

94.80%

96.80%

96.02%

87.82%

96.21%

(96.5% to 97.0%)

(92.5% to 96.5%)

(96.6% to 97.1%)

(95.74% to 96.29%)

(84.34% to 90.77%)

(95.93% to 96.47%)

    LR

95.50%

68.50%

96.40%

95.10%

61.04%

96.28%

(95.2% to 95.8%)

(64.6% to 72.2%)

(96.1% to 96.7%)

(94.79% to 95.40%)

(57.18% to 64.81%)

(96.00% to 96.55%)

Major adverse cardiac events (MACE)

    ANN classifier

95.40%

99.30%

94.50%

99.73%

93.35%

93.50%

(95.1% to 95.7%)

(99.06% to 99.5%)

(94.1% to 94.9%)

(98.53% to 99.99%)

(92.95% to 93.73%)

(93.11% to 93.87%)

    RF classifier

95.20%

99.40%

94.20%

99.97%

93.35%

94.56%

(94.9% to 95.5%)

(99.2% to 99.6%)

(93.8% to 94.6%)

(99.84% to 100.00%)

(92.95% to 93.73%)

(94.23% to 94.87%)

    LR

95.20%

99.02%

94.20%

99.78%

93.35%

94.52%

(94.8% to 95.5%)

(98.6% to 99.3%)

(93.9% to 94.6%)

(99.56% to 99.90%)

(92.95% to 93.73%)

(94.20% to 94.84%)