Skip to main content

Table 1 Final subset features selected from the original feature set. From the original feature set that consisted of temporal analysis and nonstationary analysis features, a dimension reduction technique and feature selection method were adapted for more efficient subset feature and classification accuracy

From: Exploiting temporal and nonstationary features in breathing sound analysis for multiple obstructive sleep apnea severity classification

Nonstationary analysis subset features

Rank

Base

Observation

Statistics

Sequence number

1

α

f

Maximum

40

2

α

f

Maximum

42

3

α

f

Variance

34

4

α

f

Kurtosis

8

5

f

α

Kurtosis

42

6

f

α

Maximum

24

7

f

α

Standard deviation

42

8

f

α

Variance

24

9

f

α

Median

24

10

f

α

Median

45

11

f

α

Median

46

12

f

α

Mean

7

13

f

α

Kurtosis

1

14

f

α

Kurtosis

2

15

f

α

Skewness

2

Temporal analysis subset features

Rank

Energy level transition information

16

(1 x 1) from Level 1 to Level 1

17

(3 x 4) from Level 3 to Level 4

18

(4 x 1) from Level 4 to Level 1