TABLE 1.

Prediction accuracy (mean and confidence interval). Most relevant features selected by the elastic-net logistic regression. Abbreviations: DF: Dominant frequency; SaE: Sample entropy. RHE: Relative harmonic energy.

Drug	Data	Best predictive biomarkers	Accuracy (%)
Dronedarone	ECG	DF (Leads V2–V6), SaE (Leads V2, V6)	0.72 [0.70, 0.74]
	Ionic properties	GK1, GKur, GNaK, GNa, GKr, GCaL	0.76 [0.74, 0.78]
	ECG + Ionic	CV, DF (Leads V2, V4)	0.77 [0.75, 0.79]
Flecainide	ECG	SaE (Lead V1), RHE (Lead III), SaE (Lead III)	0.69 [0.67, 0.71]
	Ionic properties	GNaK, GK1, GKs, Gto	0.71 [0.69, 0.73]
	ECG + Ionic	Gto, GK1, GNaK, GKs, SaE (Lead V1)	0.72 [0.70, 0.74]
Vernakalant	ECG	RHE (Lead V4, V5), SaE (V6)	0.68 [0.66, 0.70]
	Ionic properties	Gto, GKur, GNaK	0.70 [0.68, 0.72]
	ECG + Ionic	Gto, GKur, SaE (Lead V6), RHE (Lead V4, V5)	0.70 [0.68, 0.72]