. 2021 Oct 9;2(4):626–634. doi: 10.1093/ehjdh/ztab080

Table 2.

Heart failure prediction results

HF risk prediction method	Model inputs (‘X’ represents inputs used in corresponding method)				AUC (95% CI) on 20% hold-out test data
HF risk prediction method	ECG-AI output	ECG	ARIC variables^a	FHS variables^b	AUC (95% CI) on 20% hold-out test data
CNN (ECG-AI)		X			0.756 (0.717–0.795)
ARIC risk calculator			X		0.802 (0.750–0.850)
FHS risk calculator				X	0.778 (0.740–0.830)
Cox	X		X	X	0.818 (0.777–0.858)

ARIC, Atherosclerosis Risk in Communities; AUC, area under the receiver operating characteristic curve; BMI, body mass index; CI, confidence interval; CNN, convolutional neural network; ECG-AI, electrocardiographic artificial intelligence; FHS, Framingham Heart Study; HF, heart failure.

ARIC variables: age, gender, race, BMI, smoking status, prevalent coronary heart disease, diabetes mellitus, systolic blood pressure, heart rate.

FHS variables: age, BMI, prevalent coronary heart disease, diabetes mellitus, systolic blood pressure, left ventricular hypertrophy, valvular disease, heart rate.