Parkinson's freezing-of-gait pipeline (freeze index)

Detects freezing of gait (FOG) from a windowed spectral freeze index: the ratio of power in the freeze band (default 3-8 Hz) to power in the locomotor band (default 0.5-3 Hz) of an accelerometer signal (Moore et al. 2008; Bachlin et al. 2010). A window is flagged as FOG when the freeze index exceeds fi_threshold and the band power exceeds power_threshold (excluding rest).

Usage

pipelinePDfog(
  accel,
  sampling_rate,
  window_sec = 4,
  step_sec = 0.5,
  freeze_band = c(3, 8),
  locomotor_band = c(0.5, 3),
  fi_threshold = 2,
  power_threshold = NULL
)

Arguments

accel: Accelerometer signal (e.g. shank/trunk vertical or resultant).
sampling_rate: Sampling rate in Hz.
window_sec, step_sec: Sliding-window length and step (seconds).
freeze_band, locomotor_band: Frequency bands (Hz).
fi_threshold: Freeze-index threshold for FOG (default 2).
power_threshold: Minimum band power for a window to be considered (excludes rest). NULL uses the 10th percentile of windowed band power as a quiet-baseline floor (robust to a loud walking bout, unlike a maximum-relative gate); pass a fixed absolute value if the sensor units are known.

Value

A pd_fog_report object with a per-window windows data frame (time, freeze_index, power, fog) and the overall fog_fraction.

References

Moore ST, et al. (2008); Bachlin M, et al. (2010).