Skip to main content
. 2023 Feb 2;23(3):1625. doi: 10.3390/s23031625

Table 2.

Correctness of the Pan–Tompkins and AMPT algorithms by dataset. A1 = High Quality, A2 = Low Quality, B1 = Normal Sinus Rhythm, B2 = Arrhythmias, C = Paced Rhythm, and D = TeleHealth datasets.

Dataset Annotated Peaks Algorithm True Positives (Beats) False Positives (Beats) False Negatives (Beats) Failed Detection (Beats) Error Rate (%) Sensitivity (%) Positive Predictive Value (%) Accuracy (%)
A1 72,415 Pan–Tompkins 72,073 191 348 539 0.74 99.51 99.75 99.26
AMPT 72,267 135 148 283 0.39 99.80 99.82 99.62
A2 78,618 Pan–Tompkins 64,653 11137 14,065 25,202 32.06 80.75 85.12 74.09
AMPT 64,993 8671 13,639 22,310 28.38 82.06 86.37 78.54
B1 48,494 Pan–Tompkins 45,231 134 2988 3122 6.44 95.08 99.74 94.85
AMPT 45,301 8 3083 3091 6.37 94.97 99.98 94.96
B2 103,724 Pan–Tompkins 99,783 349 3720 4069 3.92 96.36 99.66 96.09
AMPT 100,135 144 3380 3524 3.40 96.80 99.83 96.66
C 8923 Pan–Tompkins 6684 2132 2238 4370 48.97 74.93 75.97 64.26
AMPT 8468 276 454 730 8.18 96.90 95.07 92.81
D 6708 Pan–Tompkins 3218 972 522 1494 40.10 84.92 77.56 68.91
AMPT 3394 707 334 1041 27.94 90.00 81.04 76.29