Usage

Installation

To use pyoxynet, first install it using pip:

(.venv) $ pip install pyoxynet

Note

Packages that require addition extra url cannot be installed via setuptools, which letely allows and suggests to use pip when possible. Therefore, as a work-around, TFLite is automatically installed with the following command the first time pyoxynet is imported.

The executed commend is:

(.venv) $ pip install --extra-index-url https://google-coral.github.io/py-repo/ tflite_runtime

Currently, The TFLite installation process is completed with a line command inside Python, which i know is not the best solution.

Functions

To test the presence of the package you can use the utilities.PrintHello() function:

utilities.PrintHello()

This function prints to screen.

Args:: name (str): The name to use.
Returns:: none

Every-day functions

The optimal filter adopted for the project utilities.optimal_filter(t, y, my_lambda) function:

utilities.optimal_filter(t, y, my_lambda)

A bad ass optimisation filter

Parameters:

tarray: Independent coord array
yarray: Dependent coord array
my_lambdafloat: Smoothing factor

Returns:

xarray: Filtered variable

utilities.normalize(df)

Pandas df normalisation

Parameters:: df (pd df) : input df
Returns:: result (pd df) : output df

utilities.random_walk(length=1, scale_factor=1, variation=1)

Random walk generator

Parameters:: length (int): Length of the output list scale_factor (float): Scale factor to be applied to the whole output variation (float): Local variation of the main signal with the random walk
Returns:: none

CPET data functions

utilities.load_csv_data(csv_file='data_test.csv')

Loads data from csv file (returns test data if no arguments)

Parameters:: n_inputs (int) : Number of input variables. past_points (int) : Number of past inputs in the time-series.
Returns:: df (pandas df) : Model output example

TFLite model

To test if the TFLite model has been correctly initiated you can use utilities.test_tfl_model(interpreter) function:

utilities.test_tfl_model()

Test if the model is running correclty

Parameters:: interpreter (loaded tf.lite.Interpreter) : Loaded interpreter TFLite model
Returns:: x (array) : Model output example

utilities.load_tf_model(n_inputs=7, past_points=40)

This function loads the saved tflite models.

Args:: n_inputs (int): Number of input variables. past_points (int): Number of past inputs in the time-series.
Returns:: interpreter (tflite interpreter) : handle on the TFLite interpreter

utilities.load_tf_generator()

This function loads the saved tflite generator model.

Args:: None
Returns:: generator (tflite generator) : handle on the TFLite generator

utilities.pip_install_tflite()

Makes sure TFLite is installed by executing a pip install command from the command line (sub-optimal solution)

Parameters:: none
Returns:: none

Production functions

utilities.test_pyoxynet(input_df=[], n_inputs=7, past_points=40)

Runs the pyoxynet inference

Parameters:: n_inputs (int) : Number of inputs (deafult to Oxynet configuration) past_points (int) : Number of past points in the time series (deafult to Oxynet configuration)
Returns:: x (array) : Model output example

utilities.create_probabilities(duration=600, VT1=320, VT2=460)

Creates the probabilities of being in different intensity domains

These probabilities are then sent to the CPET generator and they are used ot generate CPET vars that can replicate those probabilities

Parameters:: duration (int): Length of the test file VT1 (int): First ventilatory threshold, in time samples from the beginning of the test VT2 (int): Second ventilatory threshold, in time samples from the beginning of the test y_pm0 (double): This is used to set the first value of the probabilities. Indeed if you start from a higher VO2 you’re not likely full in moderate domain (pm<1)
Returns:: p_mF (np array): Probability of being in the moderate intensity zone (-1:1) p_hF (np array): Probability of being in the heavy intensity zone (-1:1) p_sF (np array): Probability of being in the severe intensity zone (-1:1)

utilities.generate_CPET(generator, plot=False, fitness_group=None)

Actually generates the CPET file

Parameters:: length (int): Length of the output list fitness_group (int): Fitness level: low (1), medium (2), high (3). Default to random. noise_factor (float): Noise factor for white noise. Default to random.
Returns:: df (pd df): Pandas dataframe with CPET data included and ready to be processed by the model (if needed) data (dict): Data relative to the generated CPET