Table 6.

Evaluation metrics for assessing the performance of ECG waveform delineation on data from the wearable device using the deep learning model. The table presents various statistical measures, including sensitivity, precision, F1-score, and overall accuracy, calculated for each waveform class (P-wave, QRS complex, T-wave) and the “N/W” class representing non-waveform regions. Note: The numbers represent wave counts.

Metric	N/W	P	QRS	T	Macro AVG	Micro AVG
True positive	2631	646	993	989	1314.75	1314.75
False positive	38	53	11	7	27.25	27.25
False negative	53	38	7	11	27.25	27.25
True negative	2646	4631	4357	4361	3998.75	3998.75
Precision	0.9858	0.9242	0.9890	0.9930	0.9730	0.9797
Sensitivity	0.9803	0.9444	0.9930	0.9890	0.9767	0.9797
Specificity	0.9858	0.9887	0.9975	0.9984	0.9926	0.9932
Accuracy	0.9797	0.9797	0.9797	0.9797	0.9797	0.9797
F1-measure	0.9830	0.9342	0.9910	0.9910	0.9748	0.9797