Table 4 Features extracted from the sensor data.

From: Automating sleep stage classification using wireless, wearable sensors

Sensor modality

Sampling frequency (Hz)

No. of features

Features

Accelerometer

62.5

33

Mean (x,y,z)

Minimum (x,y,z)

Maximum (x,y,z)

Range (x,y,z)

Interquartile range (x,y,z)

Standard deviation (x,y,z)

Kurtosis (x,y,z)

Root mean squared (x,y,z)

Variance (x,y,z)

Pearson’s coefficient (x,y,z)

Pearson’s p value (x,y,z)

ECG

1000

14

Mean R-R interval

Minimum R-R interval

Maximum R-R interval

Standard deviation R-R interval

RMSSD

NN50, PNN50

NN20, PNN20

VLF, LF, HF

LF/HF Ratio

Skin temperature

0.0167

4

Mean DPG

Minimum DPG

Maximum DPG

Range DPG

  1. RMSSD root mean square of successive differences; NNX number of successive R-R intervals that differ by more than X ms, PNNX ratio of NNX to total number of R-R intervals, VLF very low frequency power (activity in the 0.003–0.04 Hz frequency band); LF low frequency power (activity in the 0.04–0.15 Hz frequency band), HF high frequency power (activity in the 0.15–0.40 Hz frequency band); DPG distal-to-proximal gradient35