Detect long signal-dropout intervals — detect_gazepoint

Flags long missing, zero, constant, or low-variance runs in selected signal columns. The function is intended for quality control and audit reporting only. It does not remove samples and does not make physiological, psychological, diagnostic, or clinical claims.

Usage

detect_gazepoint_nonwear(
  data,
  signal_cols,
  group_cols = NULL,
  time_col = NULL,
  min_run_length = 10,
  zero_tolerance = 0,
  constant_tolerance = 0,
  low_variance_threshold = NULL,
  detect_missing = TRUE,
  detect_zero = TRUE,
  detect_constant = TRUE,
  detect_low_variance = TRUE
)

Arguments

data: A data frame.
signal_cols: Character vector of numeric signal columns to inspect.
group_cols: Optional character vector of grouping columns, such as participant, trial, session, or condition.
time_col: Optional numeric time column used to report interval start and end times.
min_run_length: Minimum number of consecutive samples required for a run to be reported.
zero_tolerance: Absolute tolerance used when detecting zero-valued runs.
constant_tolerance: Absolute adjacent-difference tolerance used when detecting constant runs.
low_variance_threshold: Optional standard-deviation threshold for low-variance windows. If NULL, low-variance detection is skipped.
detect_missing: Logical. If TRUE, detect non-finite runs.
detect_zero: Logical. If TRUE, detect near-zero runs.
detect_constant: Logical. If TRUE, detect constant runs.
detect_low_variance: Logical. If TRUE, detect low-variance windows when low_variance_threshold is supplied.

Value

A list with class gazepoint_nonwear_detection, containing interval and summary tables.