Table 2 Overview of the extracted features.

From: Wearable monitoring of sleep-disordered breathing: estimation of the apnea–hypopnea index using wrist-worn reflective photoplethysmography

Features type

Number of features (number of those with window shorter than literature)

Described in

Features computation window sizes (s) (number of epochs for template comparison)

HRV features

Arousal probability

10 (5)

35#,&

120#, 30&

Frequency analysis

23 (5)

34#,&,36#,37#

300#, 120&

Adapted frequency analysis

11 (5)

38#,&

300#, 120&

Detrended fluctuation analysis

5

39,40,41

360

Progressive detrended fluctuation analysis

1

42

60

Windowed detrended fluctuation analysis

1

43

360

High frequency pole analysis

4 (2)

44#,&

300#, 120&

Multi-scale entropy

20

45,46,47

540

Local phase coordination

7 (5)

48#,49#,50&

150#, 90&

Time analysis and statistics

74 (37)

34#,&

300#, 30&

Sample entropy

2 (1)

51#,&

300#, 30&

Visibility graph analysis

13

52,53,54,55

210

Hilbert transformation analysis

12 (6)

56#,&

300#, 30&

Respiratory activity features

Amplitude analysis

5

57

150

Frequency analysis

3

58

30

Similarity respiratory patterns

2

59

30 (60)

Frequency peak analysis

3

60

30

Template distance

1

58

30 (50)

Time analysis

3

61

30

Visibility graph analysis

4

52,53,54,55

150

Variance

1

60

30

Sample entropy

1

60

180

Accelerometer features

Activity counts

1

29

30

Context features

Sleep stage probabilities

4

16

30

Features coverage

1

30

  1. We calculated each feature using the methods proposed in the respective original methodological paper(s). Some of the features were calculated for different window sizes, and their number comprises both calculations (superscript symbols define the association between reference and the window size used).