Skip to main content

Table 2 The model’s performance for patient classes and transfer to an inappropriate hospital

From: Prediction of inappropriate pre-hospital transfer of patients with suspected cardiovascular emergency diseases using machine learning: a retrospective observational study

Prediction model

Count (%)

AUROC

(95% CI)

AUPRC

(95% CI)

Sensitivity (95% CI)

Specificity (95% CI)

Accuracy (95% CI)

PPV

(95% CI)

NPV

(95% CI)

Predicted cardiopulmonary resuscitation (Class 1)

525

(1.86)

0.865

(0.845, 0.885)

0.155

(0.127, 0.189)

0.802

(0.766, 0.834)

0.757

(0.752, 0.762)

0.758

(0.753, 0.763)

0.059

(0.056, 0.061)

0.995

(0.994, 0.996)

Predicted intubation (Class 2)

1,520

(5.36)

0.882

(0.871, 0.893)

0.361

(0.338, 0.386)

0.827

(0.807, 0.845)

0.778

(0.773, 0.783)

0.781

(0.776, 0.786)

0.175

(0.170, 0.179)

0.988

(0.986, 0.989)

Predicted central catheterization (Class 3)

1,025

(3.62)

0.847

(0.832, 0.862)

0.193

(0.170, 0.219)

0.789

(0.763, 0.813)

0.760

(0.755, 0.765)

0.761

(0.756, 0.766)

0.110

(0.106, 0.114)

0.990

(0.988, 0.991)

Predicted massive transfusion (Class 4)

174

(0.62)

0.740

(0.697, 0.782)

0.021

(0.007, 0.056)

0.724

(0.653, 0.785)

0.640

(0.634, 0.646)

0.641

(0.635, 0.646)

0.012

(0.011, 0.013)

0.997

(0.997, 0.998)

Predicted emergency percutaneous coronary intervention (Class 5)

2,514

(8.89)

0.929

(0.922, 0.936)

0.589

(0.569, 0.608)

0.902

(0.889, 0.913)

0.824

(0.820, 0.829)

0.831

(0.827, 0.835)

0.334

(0.327, 0.340)

0.989

(0.987, 0.990)

Predicted intensive care unit admission after ED process (Class 6)

4,437 (15.69)

0.788

(0.779, 0.796)

0.408

(0.393, 0.422)

0.740

(0.727, 0.753)

0.697

(0.692, 0.703)

0.704

(0.699, 0.710)

0.313

(0.307, 0.319)

0.935

(0.932, 0.938)

Predicted emergency operation (Class 7)

314

(1.11)

0.706

(0.673, 0.738)

0.026

(0.013, 0.050)

0.608

(0.553, 0.661)

0.713

(0.708, 0.718)

0.712

(0.707, 0.717)

0.023

(0.021, 0.025)

0.994

(0.993, 0.995)

Predicted performed magnetic resonance imaging in the ED (Class 8)

651

(2.30)

0.816

(0.796, 0.836)

0.108

(0.086, 0.134)

0.788

(0.755, 0.818)

0.741

(0.736, 0.746)

0.742

(0.737, 0.747)

0.067

(0.064, 0.070)

0.993

(0.992, 0.994)

Predicted performed echocardiography in the ED (Class 9)

851

(3.01)

0.686

(0.666, 0.706)

0.060

(0.046, 0.078)

0.622

(0.589, 0.654)

0.658

(0.653, 0.664)

0.657

(0.652, 0.663)

0.053

(0.051, 0.056)

0.982

(0.981, 0.984)

Predicted performed computed tomography angiography in the ED (Class 10)

1,536

(5.43)

0.668

(0.653, 0.683)

0.101

(0.087, 0.117)

0.689

(0.665, 0.711)

0.557

(0.551, 0.563)

0.564

(0.558, 0.570)

0.082

(0.079, 0.085)

0.969

(0.967, 0.971)

Predicted psychiatric manage in the ED (Class 11)

148

(0.52)

0.815

(0.773, 0.857)

0.024

(0.009, 0.066)

0.831

(0.763, 0.883)

0.684

(0.678, 0.689)

0.684

(0.679, 0.690)

0.014

(0.013, 0.015)

0.999

(0.998, 0.999)

Predicted admission after ED process (Class 12)

5,774 (20.42)

0.761

(0.754, 0.769)

0.434

(0.421, 0.446)

0.696

(0.684, 0.708)

0.711

(0.705, 0.717)

0.708

(0.702, 0.713)

0.382

(0.375, 0.388)

0.901

(0.898, 0.905)

Predicted discharge after ED process (Class 13)

11,664 (41.25)

0.795

(0.790, 0.800)

0.726

(0.717, 0.734)

0.714

(0.706, 0.722)

0.726

(0.719, 0.733)

0.721

(0.716, 0.726)

0.647

(0.640, 0.653)

0.784

(0.778, 0.789)

Final model for transfer to the incapable hospital

1,770

(6.26)

0.813

(0.800, 0.825)

0.286

(0.265, 0.308)

0.739

(0.718, 0.759)

0.739

(0.733, 0.744)

0.739

(0.733, 0.744)

0.159

(0.154, 0.163)

0.977

(0.975, 0.979)

  1. Note. AUROC, area under the receiver operating characteristic curve; AUPRC, area under the receiver precision-recall curve; ED, emergency department