Interval-related Analysis

This example can be referenced by citing the package.

This example shows how to use NeuroKit to analyze longer periods of data (i.e., greater than 10 seconds) such as resting state data. If you are looking to perform event-related analysis on epochs, you can refer to this NeuroKit example here.

# Load NeuroKit and other useful packages
import neurokit2 as nk
import pandas as pd

The Dataset

First, download the dataset located on the GitHub repository.

It contains 5 minutes of physiological signals recorded at a frequency of 100Hz (5 x 60 x 100 = 30000 data points).

It contains the following signals : ECG, PPG, RSP

# Get data
data = nk.data("bio_resting_5min_100hz")

This is the resting state data from 1 participant who was asked to close his/her eyes for 8 minutes, trying not to think of anything as well as not to fall asleep.

Process the Signals

In this analysis here, we will focus on extracting ECG and RSP features. To process the respective physiological signals, you can use ecg_process() and rsp_process(). You can then then visualize these signals using ecg_plot() and rsp_plot(). For the purposes of these example, we will select just 3000 datapoints (or 30s) to visualize.

Note: Do remember to specify the correct sampling_rate (in this case, to 100Hz) in which the signals were generated, in all the relevant functions.

# Process ecg
ecg_signals, info = nk.ecg_process(data["ECG"], sampling_rate=100)

nk.ecg_plot(ecg_signals[:3000], info)

# Process rsp
rsp_signals, info = nk.rsp_process(data["RSP"], sampling_rate=100)

nk.rsp_plot(rsp_signals[:3000], info)

Extract Features

Now that we have the processed signals, we can now perform the analysis using ecg_intervalrelated() and rsp_intervalrelated(). Simply provide the processed dataframe and these functions will return a dataframe of the features pertaining to the specific signal.

These features will be quite different from event-related features (See Event-related analysis example) as these signals were generated over a longer period of time. Hence, apart from the mean signal rate, variability metrices pertaining to heart rate variability (HRV) and respiratory rate variability (RRV) are also extracted here.

nk.ecg_intervalrelated(ecg_signals)

	ECG_Rate_Mean	HRV_MeanNN	HRV_SDNN	HRV_SDANN1	HRV_SDNNI1	HRV_SDANN2	HRV_SDNNI2	HRV_SDANN5	HRV_SDNNI5	HRV_RMSSD	...	HRV_SampEn	HRV_ShanEn	HRV_FuzzyEn	HRV_MSEn	HRV_CMSEn	HRV_RCMSEn	HRV_CD	HRV_HFD	HRV_KFD	HRV_LZC
0	86.392105	[[69.47563805104409]]	[[4.903604322104105]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	[[3.8837766323814966]]	...	[[1.9786368425578704]]	[[4.256941017033136]]	[[1.2686943754499123]]	[[1.404137826745609]]	[[1.469801247099329]]	[[2.572679544052144]]	[[1.562676080749381]]	[[1.8465069301766717]]	[[2.7223482404936066]]	[[0.8731238852455482]]

1 rows × 92 columns

nk.rsp_intervalrelated(rsp_signals)

	RSP_Rate_Mean	RRV_RMSSD	RRV_MeanBB	RRV_SDBB	RRV_SDSD	RRV_CVBB	RRV_CVSD	RRV_MedianBB	RRV_MadBB	RRV_MCVBB	...	RAV_Mean	RAV_SD	RAV_RMSSD	RAV_CVSD	RSP_RVT	RSP_Symmetry_PeakTrough	RSP_Symmetry_RiseDecay	RSP_Phase_Duration_Inspiration	RSP_Phase_Duration_Expiration	RSP_Phase_Duration_Ratio
0	15.860331	134.651105	384.039474	102.06409	135.551343	0.265765	0.350618	360.5	51.1497	0.141885	...	0.386252	0.201938	0.240509	0.622674	0.093298	0.617104	0.3989	0.152429	0.229395	0.664482

1 rows × 40 columns

Optional: Segmenting the Data

If you want to segment your data for analysis, such as analyzing two separate portions of your resting state data, you can simply do so by splitting the ecg_signals dataframe into epochs using epochs_create(). Using this example dataset, let’s say you want to analyze the first half and the second half of the ECG data. This means that each halved data would last for 60 x 2.5s = 150s.

In this function, we would also specify the onset of events to be at the 0th (for the first half of the data) and the 15000th datapoint (for the second half of the data), since there are 30000 data points in total.

# Half the data
epochs = nk.epochs_create(ecg_signals, 
                          events=[0, 15000], 
                          sampling_rate=100, 
                          epochs_start=0, 
                          epochs_end=150)

This returns a dictionary of 2 processed ECG dataframes, which you can then enter into ecg_intervalrelated().

# Analyze
nk.ecg_intervalrelated(epochs)

	Label	ECG_Rate_Mean	HRV_MeanNN	HRV_SDNN	HRV_SDANN1	HRV_SDNNI1	HRV_SDANN2	HRV_SDNNI2	HRV_SDANN5	HRV_SDNNI5	...	HRV_SampEn	HRV_ShanEn	HRV_FuzzyEn	HRV_MSEn	HRV_CMSEn	HRV_RCMSEn	HRV_CD	HRV_HFD	HRV_KFD	HRV_LZC
1	1	86.389814	[[69.49767441860465]]	[[5.167181138835641]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	...	[[1.2527629684953678]]	[[4.290505175709077]]	[[1.1667625041387146]]	[[1.3234302834078784]]	[[1.3340298186240032]]	[[1.5730504365078102]]	[[1.6535777692661096]]	[[1.8035044265520663]]	[[2.336829556302458]]	[[0.8288764443746864]]
2	2	86.394396	[[69.46046511627907]]	[[4.648089725942579]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	[[nan]]	...	[[1.8817856208857746]]	[[4.080221616903997]]	[[1.3473463161429124]]	[[1.7371494891482802]]	[[1.4976794053994398]]	[[1.9578060920316949]]	[[1.5369379755191364]]	[[1.8952794259141006]]	[[3.094110204759245]]	[[0.9730288694833277]]

2 rows × 93 columns

This then returns a dataframe of the analyzed features, with the rows representing the respective segmented signals. Try doing this with your own signals!