Skip to main content

Table 1 Comparison of Models in Predicting Outcomes in Patients With Heart Failure

From: Machine learning prognosis model based on patient-reported outcomes for chronic heart failure patients after discharge

 

General information

(Model 1)

CHF-PRO

(Model 2)

General information + CHF-PRO

(Model 3)

Model 3 + Parameter adjustment

(Model 4)

 

BS

AUC

P

BS

AUC

P

BS

AUC

P

BS

AUC

P

All-Cause death

          

XGBoost

0.205

0.519

(0.518,0.522)

Reference

0.237

0.601

(0.598, 0.604)

Reference

0.080

0.607

(0.595,0.608)

Reference

0.174

0.754

(0.737,0.761)

Reference

LightGBM

0.110

0.520

(0.518,0.525)

< 0.001

0.205

0.621

(0.611,0.637)

< 0.001

0.076

0.613

(0.594,0.618)

< 0.001

0.095

0.733

(0.713,0.754)

< 0.001

RF

0.106

0.547

(0.566,0.611)

< 0.001

0.152

0.590

(0.595,0.598)

< 0.001

0.091

0.608

(0.599,0.621)

< 0.001

0.371

0.709

(0.690,0.719)

< 0.001

Logistic

0.083

0.681

(0.673,0.683)

< 0.001

0.091

0.689

(0.683,0.696)

< 0.001

0.095

0.720

(0.710,0.734)

< 0.001

0.087

0.742

(0.727,0.745)

< 0.001

NB

0.080

0.500

(0.500,0.500)

< 0.001

0.894

0.514

(0.514,0.514)

< 0.001

0.894

0.514

(0.514,0.514)

< 0.001

0.356

0.658

(0.657,0.666)

< 0.001

MLP

0.364

0.509

(0.504,0.514)

< 0.001

0.392

0.678

(0.670,0.692)

< 0.001

0.413

0.645

(0.593,0.662)

< 0.001

0.288

0.746

(0.741,0.752)

< 0.001

HF readmission

          

XGBoost

0.322

0.590

(0.588,0.594)

Reference

0.415

0.535

(0.532,0.538)

Reference

0.277

0.644

(0.641,0.646)

Reference

0.235

0.718

(0.717,0.721)

Reference

LightGBM

0.288

0.548

(0.544,0.551)

< 0.001

0.405

0.519

(0.512,0.525)

< 0.001

0.254

0.610

(0.608,0.615)

< 0.001

0.231

0.704

(0.654,0.733)

< 0.001

RF

0.269

0.552

(0.540,0.556)

< 0.001

0.349

0.539

(0.541,0.544)

< 0.001

0.265

0.580

(0.564,0.604)

< 0.001

0.216

0.707

(0.702,0.710)

< 0.001

Logistic

0.458

0.639

(0.636,0.642)

< 0.001

0.278

0.585

(0.580,0.586)

< 0.001

0.405

0.665

(0.661,0.667)

< 0.001

0.307

0.693

(0.686,0.701)

< 0.001

NB

0.261

0.500

(0.500,0.500)

< 0.001

0.265

0.508

(0.509,0.512)

< 0.001

0.261

0.498

(0.498,0.498)

< 0.001

0.390

0.673

(0.665,0.685)

< 0.001

MLP

0.409

0.585

(0.580,0.594)

< 0.001

0.485

0.562

(0.551,0.579)

< 0.001

0.390

0.615

(0.612,0.619)

< 0.001

0.220

0.690

(0.640,0.712)

< 0.001

MACEs

            

XGBoost

0.405

0.527

(0.523,0.531)

Reference

0.439

0.540

(0.538,0.544)

Reference

0.318

0.600

(0.588,0.597)

Reference

0.364

0.670

(0.595,0.710)

Reference

LightGBM

0.443

0.515

(0.512,0.521)

< 0.001

0.420

0.519

(0.508,0.530)

< 0.001

0.367

0.580

(0.579,0.607)

< 0.001

0.348

0.620

(0.594,0.644)

< 0.001

RF

0.383

0.527

(0.525,0.530)

> 0.999

0.410

0.540

(0.545,0.546)

> 0.999

0.383

0.552

(0.540,0.557)

< 0.001

0.356

0.666

(0.641,0.680)

0.003

Logistic

0.470

0.593

(0.580,0.611)

< 0.001

0.358

0.607

(0.605,0.609)

< 0.001

0.424

0.613

(0.597,0.622)

< 0.001

0.402

0.629

(0.612,0.645)

< 0.001

NB

0.337

0.500

(0.500,0.500)

< 0.001

0.337

0.497

(0.499,0.500)

< 0.001

0.337

0.497

(0.497,0.497)

< 0.001

0.455

0.657

(0.646,0.669)

< 0.001

MLP

0.576

0.530

(0.519,0.547)

< 0.001

0.394

0.619

(0.605,0.624)

< 0.001

0.394

0.575

(0.570,0.578)

< 0.001

0.284

0.670

(0.647,0.692)

> 0.999

  1. AUC, area under curve; BS, brier score; CHF-PRO, chronic heart failure - patient reported outcome; LightGBM, light gradient boosting machine; MACEs, major adverse cardiovascular events; MLP, multilayer perceptron; NB, naive bayes; RF, random forest; XGBoost, extreme gradient boosting
Back to article page

Health and Quality of Life Outcomes

ISSN: 1477-7525

Contact us