GaitPy

Gait feature extraction and bout classification from single accelerometer in the lumbar location. This class includes functions for:

• Continuous wavelet based method of gait kinematic feature extraction.

• Machine learning based method of bout classification.

• Visualizing results.

Parameters:

data: str or pandas.core.frame.DataFrame

• Option 1: Pandas dataframe containing unix timestamp column and vertical acceleration data during gait, both of type float

• Option 2: File path of .csv file containing timestamp column and vertical acceleration data during gait. One column should contain unix timestamps of type float (by default gaitpy will assume the column title is ‘timestamps’ with units in milliseconds). A second column should be vertical acceleration of type float (by default gaitpy will assume the column title is ‘y’ with units in m/s^2).

sample_rate: int or float

Sampling rate of accelerometer data in Hertz.

v_acc_col_name: str

Column name of the vertical acceleration data (‘y’ by default)

ts_col_name: str

Column name of the timestamps (‘timestamps’ by default)

v_acc_units: str

Units of vertical acceleration data (‘m/s^2’ by default). Options:

• ‘m/s^2’ = meters per second squared

• ‘g’ = standard gravity

ts_units: str

Units of timestamps (‘ms’ by default). Options:

• ‘s’ = seconds

• ‘ms’ = milli-seconds

• ‘us’ = microseconds

flip: bool

Boolean specifying whether to flip vertical acceleration data before analysis (False by default). Algorithm assumes that baseline vertical acceleration data is at -9.8 m/s^2 or -1g. (ie. If baseline data in vertical direction is 1g, set ‘flip’ argument to True)

• classify_bouts
• extract_features
• plot_contacts