Table 3 In-data-set F1-score, in percentage points, on validation data sets.

From: Robust screening of atrial fibrillation with distribution classification

Algorithm

DiagnoStick

SPH

CinC

XQRS34

\(\mathbf {75.83}\ (\pm 2.23)\)

\(\mathbf {85.05}\ (\pm 1.24)\)

\(\mathbf {69.23}\ (\pm 2.75)\)

Christov35

\(52.81\ (\pm 7.08)\)

\(78.93\ (\pm 1.47)\)

\(55.29\ (\pm 7.60)\)

Elgendi et al.36

\(63.04\ (\pm 1.97)\)

\(74.05\ (\pm 2.40)\)

\(20.16\ (\pm 4.99)\)

Hamilton37

\(73.26\ (\pm 5.45)\)

\(76.26\ (\pm 1.73)\)

\(63.90\ (\pm 4.05)\)

Rodrigues et al.38

\(73.80\ (\pm 2.03)\)

\(84.06\ (\pm 1.16)\)

\(67.19\ (\pm 2.54)\)

Zong et al.39

\(18.80\ (\pm 4.94)\)

\(71.79\ (\pm 1.41)\)

\(52.92\ (\pm 2.62)\)

Makowski et al.40

\(73.54\ (\pm 3.77)\)

\(84.99\ (\pm 2.08)\)

\(64.51\ (\pm 4.46)\)

Pan and Tompkins41

\(71.68\ (\pm 3.35)\)

\(80.33\ (\pm 1.12)\)

\(65.02\ (\pm 4.53)\)

  1. Bold font indicates the best of each column. The scores are the means over the held-out fold during cross-validation, and the value in parentheses is the variance over that fold. The main results of Table 2 use the XQRS algorithm.