Table 2 (Performance indexes) Scores of our DNN are compared on the test set with the average performance of: (i) 4th year cardiology resident (cardio.); (ii) 3rd year emergency resident (emerg.); and (iii) 5th year medical students (stud.).

From: Automatic diagnosis of the 12-lead ECG using a deep neural network

 

Precision (PPV)

Recall (Sensitivity)

Specificity

F1 score

 

DNN

cardio.

emerg.

stud.

DNN

cardio.

emerg.

stud.

DNN

cardio.

emerg.

stud.

DNN

cardio.

emerg.

stud.

1dAVb

0.867

0.905

0.639

0.605

0.929

0.679

0.821

0.929

0.995

0.997

0.984

0.979

0.897

0.776

0.719

0.732

RBBB

0.895

0.868

0.963

0.914

1.000

0.971

0.765

0.941

0.995

0.994

0.999

0.996

0.944

0.917

0.852

0.928

LBBB

1.000

1.000

0.963

0.931

1.000

0.900

0.867

0.900

1.000

1.000

0.999

0.997

1.000

0.947

0.912

0.915

SB

0.833

0.833

0.824

0.750

0.938

0.938

0.875

0.750

0.996

0.996

0.996

0.995

0.882

0.882

0.848

0.750

AF

1.000

0.769

0.800

0.571

0.769

0.769

0.615

0.923

1.000

0.996

0.998

0.989

0.870

0.769

0.696

0.706

ST

0.947

0.968

0.946

0.912

0.973

0.811

0.946

0.838

0.997

0.999

0.997

0.996

0.960

0.882

0.946

0.873

  1. PPV positive predictive value. The bold values represent the best scores.