ada.long.nonparametric#

Classes#

`DFA`	A class for estimating Hurst exponent of the data using Detrended Fluctuation Analysis.
`DFAResults`	A container for results of fitting DFA model to the actigraphic data. Do not construct manually! The following parameters are available in the fit_params:
`SimpleNonparametric`	A class for estimating basic metrics reflecting 24 hour periodicity in the data. These are:
`SimpleNonparametricResults`	Container for simple nonparametric metrics. Do not construct manually! Following fields are available in fit params (coresponding to the fitted metrics):

Module Contents#

class DFA(n_windows=200, scale_range=(1 / 6, 48), trend_degree=1, n_crossovers=1)[source]#

A class for estimating Hurst exponent of the data using Detrended Fluctuation Analysis.

n_windows#

Number of scales that will be taken into account during computation. Defaults to 200.

Type:

int

Parameters:

n_windows (int)
scale_range (Tuple[float, float])
trend_degree (int)
n_crossovers (int | None)

scale_range#

Range of scales in hours. Defaults to (1 / 6, 48).

Type:

Tuple[float, float]

Parameters:

n_windows (int)
scale_range (Tuple[float, float])
trend_degree (int)
n_crossovers (int | None)

trend_degree#

Degree of polynomial function describing trend in the integrated data (degree of trend in original data is equal to degree of trend in integrated data - 1). Defaults to 1.

Type:

int

Parameters:

n_windows (int)
scale_range (Tuple[float, float])
trend_degree (int)
n_crossovers (int | None)

n_crossovers#

Number of crossover points expected in the fluctuation graphs, assuming that the data is described by n_crossovers + 1 fractals. If None, optimal number of crossovers in the range 0-10 will be found using Bayes Inforation Crieteria. Defaults to 1.

Type:

int | None

Parameters:

n_windows (int)
scale_range (Tuple[float, float])
trend_degree (int)
n_crossovers (int | None)

fit(data, ch_name=None)[source]#

Fits the specified by the object fractal model to provided data.

Parameters:

data (_ActiData) – Actigraphic data to be fitted with cosinor method.
ch_name (str | None, optional) – Channel to which model will be fitted. If None, it will be to_score in case of Raw or Epoched data and score in case of ScoredShort data. Defaults to None.

Returns:

Object containing fitting results.

Return type:

DFAResults

class DFAResults(scales, fluctuations, exponent, scale_range, crossovers, degree, id)[source]#

A container for results of fitting DFA model to the actigraphic data. Do not construct manually! The following parameters are available in the fit_params:

exponent: the exponent of the whole data over all fluctuation range;
exponent error: the error of exponent estimation;
scale range: range of scales used during the fitting;
segments: segments in which the data was divided;
trend degree: trend degree used during the detrending prior to fitting model.

Parameters:

scales (numpy.ndarray)
fluctuations (numpy.ndarray)
exponent (Tuple[float, float])
scale_range (Tuple[float, float])
crossovers (Tuple[dict | None, numpy.ndarray])
degree (int)
id (str)

export(out_path)[source]#

Save data (both parameters of model and fluctuations data) to compressed generic file.

Parameters:: out_path (str) – Path to the out file.

static load_file(path)[source]#

Load model fitting results from .ada.long file (created via export method).

Parameters:: path (str) – Path to file.
Returns:: Object containing fitted model.
Return type:: DFAResults

plot(out_path=None)[source]#

Plot the semi-logarithmic fluctuations plot.

Parameters:: out_path (str | None, optional) – Path to save the figure. If None, it will be shown in interactive mode. Defaults to None.

save_csv(out_path)[source]#

Save parameters of fitted model to human-readable csv.

Parameters:: out_path (str) – Path to the out file.

property crossovers: Tuple[list[float], list[Tuple[float, float]]]#

List with positions of crossovers (in hours) and list of their confidence intervals.

Return type:: Tuple[list[float], list[Tuple[float, float]]]

property exponents: Tuple[list[float], list[float]]#

List with exponents and list of their errors.

Return type:: Tuple[list[float], list[float]]

property fit_params: dict#

Parameters of the fitted model.

Return type:: dict

property fluctuations: numpy.ndarray#

Array of obtained fluctuations at different scales.

Return type:: numpy.ndarray

property id: str#

ID of a recording to which long estimate was fit.

Return type:: str

property piecewise_fit: numpy.ndarray#

Array with reconstruction of fitted piecewise linear function.

Return type:: numpy.ndarray

property scales: numpy.ndarray#

Array of scale for which fluctuations were computed.

Return type:: numpy.ndarray

class SimpleNonparametric[source]#

A class for estimating basic metrics reflecting 24 hour periodicity in the data. These are:

Interdaily stability (IS): reflects how similar days are to each other in 24 hour windows. Contained in [0, 1] with higher values meaning higher regularity.
Intradaily variability (IV): reflects variability of activity levels during the day; higher values mean higher fragmentation (potential circadian disturbances).
M10: mean activity in the most active 10 hours during each day.
L5: mean activity in the least active 5 hours during each day.

For details see Witting W, Kwa IH, Eikelenboom P, Mirmiran M, Swaab DF. Alterations in the circadian rest-activity rhythm in aging and Alzheimer’s disease. Biol Psychiatry. 1990 Mar 15;27(6):563-72.

fit(data, ch_name=None)[source]#

Computes basic nonparametric metrics estimating 24 hour periodicity in the data.

Parameters:

data (_ActiData) – Input data.
ch_name (str | None, optional) – Channel on which computation will be done. If None, default will be used. Defaults to None.

Returns:

Container with all the results.

Return type:

SimpleNonparametricResults

window_size = 60#

class SimpleNonparametricResults(interdaily_stability, intradaily_variability, m10, l5, id)[source]#

Container for simple nonparametric metrics. Do not construct manually! Following fields are available in fit params (coresponding to the fitted metrics):

export(out_path)[source]#

Save data to compressed generic file.

Parameters:: out_path (str) – Path to the out file.

static load_file(path)[source]#

Load model fitting results from .ada.long file (created via export method).

Parameters:: path (str) – Path to file.
Returns:: Object containing fitted model.
Return type:: SimpleNonparametricResults

plot(out_path=None)[source]#

There is absolutely nothing to plot when it comes to this metrics.

Parameters:: out_path (str | None)

save_csv(out_path)[source]#

Save parameters of fitted model to human-readable csv.

Parameters:: out_path (str) – Path to the out file.

property fit_params: dict#

Parameters of the fitted model.

Return type:: dict

property id: str#

ID of a recording to which long estimate was fit.

Return type:: str