Skip to main content
. Author manuscript; available in PMC: 2019 Oct 9.
Published in final edited form as: Nat Med. 2019 Jan 7;25(1):65–69. doi: 10.1038/s41591-018-0268-3

Table 1:

Diagnostic performance of the deep neural network and averaged individual cardiologists compared to the cardiologist committee consensus (n=328). a, Deep Neural Network algorithm area under the receiver operating characteristic curve (AUC) compared to the cardiologist committee consensus. b, Deep Neural Network algorithm and averaged individual cardiologist F1 scores compared to the cardiologist committee consensus.

Table 1a
Sequence* AUC (95% CI) Set** AUC (95% CI)
Atrial Fibrillation & Flutter 0.973 (0.966–0.980) 0.965 (0.932–0.998)
Atrio-ventricular Block 0.988 (0.983–0.993) 0.981 (0.953–1.000)
Bigeminy 0.997 (0.991–1.000) 0.996 (0.976–1.000)
Ectopic Atrial Rhythm 0.913 (0.889–0.937) 0.940 (0.870–1.000)
Idioventricular Rhythm 0.995 (0.989–1.000) 0.987 (0.959–1.000)
Junctional Rhythm 0.987 (0.980–0.993) 0.979 (0.946–1.000)
Noise 0.981 (0.973–0.989) 0.947 (0.898–0.996)
Sinus Rhythm 0.975 (0.971–0.979) 0.987 (0.976–0.998)
Supraventricular Tachycardia 0.973 (0.960–0.985) 0.953 (0.903–1.000)
Trigeminy 0.998 (0.995–1.000) 0.997 (0.979–1.000)
Ventricular Tachycardia 0.995 (0.980–1.000) 0.980 (0.934–1.000)
Wenckebach 0.978 (0.967–0.989) 0.977 (0.938–1.000)

Average 0.978 0.977
Table 1b
Algorithm
Sequence-level*
F1
Algorithm
Set-level**
F1
Average
Cardiologist
Sequence-level F1
Average
Cardiologist
Set-level F1
Atrial Fibrillation & Flutter 0.801 0.831 0.677 0.686
Atrio-ventricular Block 0.828 0.808 0.772 0.761
Bigeminy 0.847 0.870 0.842 0.853
Ectopic Atrial Rhythm 0.541 0.596 0.482 0.536
Idioventricular Rhythm 0.761 0.818 0.632 0.720
Junctional Rhythm 0.664 0.789 0.692 0.679
Noise 0.844 0.761 0.768 0.685
Sinus Rhythm 0.887 0.933 0.852 0.910
Supraventricular Tachycardia 0.488 0.693 0.451 0.564
Trigeminy 0.907 0.864 0.842 0.812
Ventricular Tachycardia 0.541 0.681 0.566 0.769
Wenckebach 0.702 0.780 0.591 0.738

Average 0.807 0.837 0.753 0.780
*

Sequence-level: describes algorithm predictions that are made once every 256 input samples (approximately every ~1.3 seconds) and are compared against the gold-standard committee consensus at the same intervals.

**

Set-level: describes the unique set of algorithm predictions that are present in the 30-second record. Sequence AUC prediction n=7544; Set AUC prediction n=328.