Skip to contents

Downsample 5-Minute CGM Data to 15-Minute Intervals

Source: R/interval_down.R
interval_down.Rd

Converts Dexcom-style 5-minute CGM data to clock-aligned 15-minute intervals. Each subject is grouped by actual time, not by row position, so a missing 5-minute record does not shift later intervals. Each aggregate is labeled at the right boundary of its 15-minute interval: :15, :30, :45, or :00 past the hour.

Usage

interval_down(df, n_observed = FALSE)

Arguments

df

A data frame with columns id, time (POSIXct), and gl (numeric glucose values).

n_observed

Logical. If TRUE, include an n_observed column with the number of glucose readings contributing to each result. Defaults to FALSE.

Value

A data frame with id, clock-aligned time, and averaged gl columns. If n_observed = TRUE, it also includes n_observed. The time column retains the input POSIXct time zone.

Details

The glucose value is the mean of all non-missing readings in the interval. A single available value is retained and assigned to the interval-end output time. Intervals with no available glucose values are omitted. n_observed = TRUE adds a column recording how many glucose readings contributed to each result. Rows must be ordered by subject and time before calling this function.

Examples

dexcom <- data.frame(
  id = "participant_1",
  time = as.POSIXct(
    c("2024-01-01 00:00:00", "2024-01-01 00:05:00",
      "2024-01-01 00:10:00", "2024-01-01 00:15:00",
      "2024-01-01 00:20:00", "2024-01-01 00:25:00"),
    tz = "UTC"
  ),
  gl = c(100, 110, 120, 130, 140, 150)
)

interval_down(dexcom)
#>              id                time  gl
#> 1 participant_1 2024-01-01 00:15:00 110
#> 2 participant_1 2024-01-01 00:30:00 140