pyheartlib.beat_info

Module Contents

Classes

BeatInfo

Provides information and features for a single beat.
class pyheartlib.beat_info.BeatInfo(beat_loc=10, fs=360, in_ms=True)

Provides information and features for a single beat.

Parameters:

  • beat_loc (int) – Index of beat in the R-peaks locations list, considering preceding R-peak locations and subsequent R-peak locations.

  • fs (int, optional) – Sampling rate, by default 360

  • in_ms (bool, optional) – Whether to calculate RR-intervals in time (miliseconds) or samples, by default True

beat_loc

Index of beat in the R-peaks locations list, considering preceding R-peak locations and subsequent R-peak locations.

Type:

int

fs

Sampling rate, by default 360

Type:

int, optional

in_ms

Whether to calculate RR-intervals in time (miliseconds) or samples, by default True

Type:

bool, optional

input_waveform

Input waveform, that is not trimmed (len_waveform, #channels).

Type:

numpy.ndarray

bwaveform

Segmented beat waveform. Selected as a segment of the input waveform (len_waveform, #channels).

Type:

numpy.ndarray

start_idx

The index of the first sample (onset) of the waveform on the original signal.

Type:

int

rri

List of RR-intervals, in miliseconds if the in_ms parameter has been set to True.

Type:

list

rri_smpl

RR-intervals in samples.

Type:

list

sdrri

Successive RR interval (RRI) differences.

Type:

list

avail_features

List of features names (strings) that already have a definition (predefined features).

Type:

list

features

Dictionary containing computed features. Keys are feature names and values are features values.

Type:

dict

Notes

BeatInfo includes some predefined features, however custom features can also be defined. Each custom feature definition must adhere to this syntax and its name must start with F_:

>>> def F_new_feature(self):
>>>    return return_result

The return_result must be one of the following:

  1. A real number.

2) A dictionary such as {“Feature_1”: value_1, “Feature_2”: value_2}. Each value can be a real number. Alternatively, each value can be a one-dimensional array, list, or tuple. In this case, their elements must correspond to the channels.

3) Tuple or one-dimensional NumPy array. Their elements must correspond to the channels. For example, the return value of F_feat_new() as a tuple (element1, element2) will produce F_feat_new(CH1) for element1 and F_feat_new(CH2) for element2. The order of channels is determined by the CHANNEL field in the config.yaml file.

  1. A list. An output as a list will be a list.

The custom features can be added to the beatinfo object by using the add_features() method, the list of all available features can be obtained using the available_features() method, and to select desired features for computation the select_features() method can be used.

__call__(data)
Parameters:

data (dict) –

A dict containing data about the beat with keys:

’waveform’numpy.ndarray

Waveform (len_waveform, #channels).

’rpeak_locs’list

A list containing locations of rpeaks.

’rec_id’int

Record id of the original signal which the excerpt is extracted from.

’start_idx’int

Index of beginnig of the segmented waveform in the original signal.

’label’str

Type of the beat.

available_features()

Returns available features that can be computed.

Returns:

List containing features names.

Return type:

list

add_features(new_features)

Adds new features.

Parameters:

new_features (list) – Names of new features in a list. Such as: [new_feature_1, new_feature_2]

select_features(features)

Select desired features for computation.

Parameters:

features (list) – List of features names (strings).

compute_features()

Computes features.

Returns:

Dictionary with keys(strings) equal to feature names and dict values equal to features values.

Return type:

dict

get_beat_waveform(win=[-0.35, 0.65])

Segment beat waveform according to the pre and post RR-intervals.

get_rris(in_ms)

Compute RR-intervals.

Parameters:

in_ms (bool) – If True result will be in time(miliseconds), otherwise in samples.

Returns:

Contains RR-intervals.

Return type:

list

get_sdrri()

Computes successive differences of rri.

Returns:

Contains successive differences of rri.

Return type:

list

F_post_rri()

Post RR interval.

F_pre_rri()

Pre RR interval.

F_ratio_post_pre()

Ratio of post and pre RR-intervals.

F_diff_post_pre()

Difference of post and pre RR-intervals.

F_diff_post_pre_nr()

Normalized difference of post and pre RR-intervals.

F_median_rri()

Median of RR-intervals.

F_mean_pre_rri()

Mean of pre RR-intervals.

F_rms_rri()

RMS of RR-intervals.

F_std_rri()

STD of RR-intervals.

F_ratio_pre_rms()

Ratio of pre RR interval to RMS of RR-intervals.

F_ratio_post_rms()

Ratio of post RR interval to RMS of RR-intervals.

F_diff_pre_avg_nr()

Difference of pre RR interval and median of RR-intervals, normalized by their RMS.

F_diff_post_avg_nr()

Difference of post RR interval and median of RR-intervals, normalized by their RMS.

F_compensate_ratio()

Compensation ratio.

F_compensate_diff_nr()

Compensation diff normalized by RMS.

F_heart_rate()

Heart rate.

F_mean_sdrri()

Mean of successive RRI differences.

F_absmean_sdrri()

Mean of absolute successive RRI differences.

F_mean_pre_sdrri()

Mean of pre successive RRI differences.

F_rms_sdrri()

RMS of successive RRI differences.

F_std_sdrri()

STD of successive RRI differences.

reported_rpeak()
F_beat_max()

Max value of heartbeat waveform.

F_beat_min()

Min value of heartbeat waveform.

F_maxmin_diff()

Difference of max and min values of heartbeat waveform.

F_maxmin_diff_norm()

Difference of max and min values of heartbeat waveform, normalized by waveform RMS.

F_beat_mean()

Mean of heartbeat waveform.

F_beat_std()

STD of heartbeat waveform.

F_beat_skewness()

Skewness of heartbeat waveform.

F_beat_kurtosis()

Kurtosis of heartbeat waveform.

F_beat_rms()

RMS of heartbeat waveform.

F_nsampels(n_samples=10)

Gives equally spaced samples of the trimmed heartbeat waveform.

Parameters:

n_samples (int, optional) – Number of samples, by default 10

Returns:

Waveform sample values.

Return type:

dict

F_fft_features()

Returns FFT of each heartbeat trimmed waveform.

Return type:

dict