From: Machine learning methods for the study of cybersickness: a systematic review
Author | Biosignal | Preprocessing |
---|---|---|
Preprocessing methods | ||
Nam et al. [12] | EEG, EOG, ECG, finger tip skin temperature, PPG, skin conductance | Power band extraction, standard deviation of EOG, mean R–R of ECG, mean and standard deviation of fingertip skin temperature, PPG and skin conductivity. Data segments for all variables calculated in period 3 (30 s after to the end of VR immersion) ratioed to period 1 and 2 (1 min before VR immersion and 30 s after) |
Yu et al. [13] | EEG | 1–50 Hz high and low pass filter, 250 Hz down sampling, ICA, component clustering, FFT and conversion to decibel power |
Wei et al. [16] | EEG | 1–50 Hz high and low pass filter, 250 Hz down sampling, ICA, component clustering, FFT and conversion to decibel power |
Wei et al. [14] | EEG | 1–50 Hz high and low pass filter, 250 Hz down sampling, ICA, component clustering, FFT and conversion to decibel power |
Ko et al. [15] | EEG | 1–50 Hz high and low pass filter, 250 Hz down sampling, ICA, component clustering, FFT and conversion to decibel power |
Lin et al. [17] | EEG | 1–50 Hz high and low pass filter, 250 Hz down sampling, ICA, component clustering, FFT for PSD and subsequent conversion to decibel power |
Ko et al. [18] | EEG | 1–50 Hz high and low pass filter, 250 Hz down sampling, ICA, component clustering, FFT for PSD and conversion to decibel power |
Lin et al. [19] | EEG | 1–50 Hz high and low pass filter, 250 Hz down sampling, ICA, component clustering, FFT for PSD and conversion to decibel power |
Dennison et al. [29] | ECG, EGG, EOG, blink rate, PPG, breathing rate, GSR | ECG bandpass filter 0.5–30 Hz, EGG bandpass filter 0.005–2 Hz and FFT with Hamming window, percentage band power for tachygastric and bradygastric activity, respiration bandpass filter 0.1–1 Hz, PPG bandpass filter 0.1–10 Hz, EOG bandpass filter 0.1–5 Hz, baseline normalization for skin conductivity, standard deviation of yaw, pitch and roll head rotation in degrees |
Pane et al. [26] | EEG | FIR bandpass 1–40 Hz, ICA, ratio logarithmic of PSD (percentage power), change in percentage power pre-stimuli to post stimuli (percentage change) Daubechies 4 wavelet (db4) function |
Mawalid et al. [21] | EEG | ICA, Chebyshev bandpass filter type II |
Khoirunnisaa et al. [20] | EEG | FIR bandpass 1–40 Hz, ICA, Discrete Wavelet transform, Welch's method for PSD |
Dennison et al. [25] | EEG, ECG, EOG, blink rate, breathing rate, EGG, postural sway, head movement | ECG bandpassfilter 0.5–30 Hz, EEG bandpass filter 0.1–30 Hz, data interpolation from other channels after manual artifact removal, ICA, FFT, EOG bandpass filter 0.1–5 Hz, EGG bandpass filter 0.005–2 Hz, FFT with Hamming window, percentage band power for tachygastric and bradygastric activity, respiration bandpass filter 0.1–1 Hz, standard deviation of yaw, pitch and roll rotation degrees, average and standard deviations in weight changes for postural sway. Any missing data replaced and standardized across features |
Wang et al. [34] | Postural sway | – |
Garcia-Agundez et al. [28] | ECG, EOG, blink rate, breathing rate, GSR | Mean and standard deviation on game content vectors |
Jeong et al. [22] | EEG | 4–45 Hz automatic filter. Data sets created based on 4 custom signal quality weightings, min max normalization/standardization |
Li et al. [35] | EEG, postural sway, head body movement | Channel integration, paired interception, simultaneous artifact removal, FFT for PSD |
Kim et al. [42] | EEG | Bandpass filter 0.3–100 Hz, notch filter at 60 Hz, FFT applied through a sliding Hann window. EEG Data transformed into a 8 channel stacked spectogram |
Liao et al. [27] | EEG | FFT for PSD |
Li et al. [23] | EEG | Elliptical pass band filter 0.5–30 Hz, Fourier transform, 7 level WPT |
Lee and Alamaniotis [43] | EEG | 256 Hz down sampling |
Islam et al. [30] | ECG, breathing rate, GSR | z-score removal of outliers |
Islam et al. [31] | ECG, breathing rate, GSR | z-score removal of outliers, 1 Hz down sampling, min–max normalization |
Martin et al. [33] | BVP, EDA | BVP inter-beat interval extraction bandpass filter 0.66–3.33 Hz, frequency and time domain feature computation, EDA tonic and phasic computation |
Recenti et al. [24] | EEG, EMG, heart rate | 0.1–40 Hz high pass and low pass filter, 300 microvolts upper limit, common average reference, interpolation for removed channels, baseline correction, DC offset correction, Welch's method for PSD, relative power averaged across all channels |
Oh and Kim [32] | BVP, respiratory signal | Exclusions of data samples after inspection for artifacts |