Detect candidate glucose peaks — detect_candidate

Identifies candidate peaks in a smoothed glucose trace using dynamically adjusted prominence and minimum-distance criteria, restricted to values that exceed the corrected baseline by a minimum rise and an absolute floor.

Usage

detect_candidate_peaks(
  glucose,
  baseline,
  glucose_sd,
  min_rise = 10,
  abs_floor = 70
)

Arguments

glucose: Numeric vector, smoothed glucose trace (see sg_smooth()).
baseline: Numeric scalar, corrected baseline (see correct_baseline()).
glucose_sd: Numeric scalar, glucose standard deviation used to set dynamic prominence/distance thresholds.
min_rise: Numeric, minimum rise above baseline required for a peak to be considered valid. Default 10 (mg/dL).
abs_floor: Numeric, absolute minimum glucose value for a valid peak. Default 70 (mg/dL).

Value

A data frame of candidate peaks with columns index, time (if timestamps are supplied downstream), and value.

Details

Thresholds follow the manuscript pipeline:

Prominence: max(2, glucose_sd) mg/dL
Distance: 4 points if glucose_sd < 10, else 5 points
Validation tolerance: max(5, glucose_sd) mg/dL