Table 4.

Detailed description of the set of ECG features used for the signal quality assessment.

Category	Name	Description
Statistical Features	mean	1st central moment over valid ECG signal segment
	median	Median value over valid ECG signal segment
	standard deviation	2nd central moment over valid ECG signal segment
	skewness	3rd central moment over valid ECG signal segment
	kurtosis	4th central moment over valid ECG signal segment
Time-domain Features	mean HR	Average heart rate calculated from identified RR-intervals
	SDNN	Standard deviation of RR-intervals over 1-minute recording [35]
	coverage	Segment length $T_{\mathrm{seg}}$ of valid ECG tracing (cf. Section 3.2.1)
	eCDF consistency	Mean value of eCDF correlation coefficient [36]
	Signal consistency	DTW Distance between first half and second half of ECG recording [37]
Frequency-domain Features	FFT consistency	Mean value of FFT correlation coefficient
	basSQI	Signal quality index (SQI) for ECG baseline wander according to [15]
	qrsSQI	SQI for QRS complex power according to [15]
	pliSQI	SQI for power-line interference according to [38]
	SNR	Signal-to-noise ratio according to [38]
	mean SQI	Mean SQI: (basSQI + qrsSQI + pliSQI)/3 [38]
Morphology-related Features	mean R-peak amplitude	Average R-peak amplitude from peak detection
	mean template STD	Average standard deviation from mean morphology
	Euclidean distance	Euclidean distance between mean morphology and optimal signal prototype
	Cityblock distance	One-norm between mean morphology and optimal signal prototype [39]
	DTW distance	DTW distance between mean morphology and optimal signal prototype [37]
	Correlation Coefficient	Correlation coefficient between mean morphology and optimal signal prototype