wristsed provides analysis-ready NHANES 2011--2014 wrist accelerometer sedentary-time datasets. The package contains derived epoch-level and participant-level Parquet data products based on CHAP Wrist posture predictions combined with either SWaN sleep/wake/non-wear classifications or aligned NHANES minute-level sleep states.

The package is intended for researchers who want to analyze derived sedentary behavior summaries without rebuilding the full raw accelerometer processing pipeline.

Install the package from GitHub:

# install.packages("remotes")
remotes::install_github("codoom1/wristsed")

Load the package:

library(wristsed)

The package depends on arrow, dplyr, and utils. These dependencies are declared in the package DESCRIPTION file and are installed by standard R package installation tools.

wristsed includes epoch-level and participant-level sedentary-time datasets.

The primary analysis dataset uses:

sleep_method = "swan"
epoch = "1h"

The sensitivity dataset uses:

sleep_method = "nhanes"

Sleep/wake methods:

• swan: primary analysis dataset using CHAP Wrist posture predictions combined with SWaN sleep/wake/non-wear states.
• nhanes: sensitivity dataset using CHAP Wrist posture summaries combined with aligned NHANES 1-minute sleep states.

List packaged data products:

available_datasets()

Read the default 1-hour SWaN epoch-level dataset:

epoch_1h <- get_epoch_data()
head(epoch_1h)

Read participant-level summaries:

participant <- get_participant_data()
head(participant)

List participants available across both sleep/wake methods:

participants <- available_participants(sleep_method = "all")
head(participants)

Read the primary 1-hour epoch-level dataset:

epoch_1h <- get_epoch_data(
  epoch = "1h",
  sleep_method = "swan"
)

Read the primary participant-level dataset:

participant <- get_participant_data(
  sleep_method = "swan",
  epoch = "1h"
)

Read the NHANES sleep-state sensitivity dataset:

epoch_nhanes <- get_epoch_data(
  epoch = "1h",
  sleep_method = "nhanes"
)

By default, the data accessors collect data into memory as a data frame. For larger workflows, set collect = FALSE to return an Arrow Dataset. This allows users to filter or select before collecting rows into memory.

library(dplyr)

epoch_ds <- get_epoch_data(
  epoch = "10m",
  sleep_method = "swan",
  collect = FALSE
)

morning_weekdays <- epoch_ds |>
  filter(DayType == "Weekday", epoch_start_minute < 12 * 60) |>
  select(ID, Dataset, Day, Epoch, percent_waking_sedentary) |>
  collect()

Epoch-level datasets include:

• ID, Dataset, Day, DayType, and Epoch
• epoch_level, epoch_start_minute, and epoch_duration_minutes
• sleep_method
• percent_sedentary
• percent_waking_sedentary
• percent_sleep_nonwear
• percent_wake
• percent_missing_sleep
• source_coverage

Participant-level datasets include:

• ID, Dataset, sleep_method, and epoch_level
• sleep_nonwear_cutoff
• no_of_valid_days
• est_sedentary_hours
• est_waking_hours
• est_sleep_nonwear_hours
• within-participant SD and SE columns for daily estimates
• NHANES survey variables: Exam_weight_recal, PSU, and Stratum

Use participant-level datasets for nationally representative participant-level analyses because NHANES survey weights are assigned to participants, not epochs.

Detailed data dictionaries are included with the package and can be accessed directly from R.

View the epoch-level dictionary in the console:

view_data_dictionary("epoch")

View the participant-level dictionary:

view_data_dictionary("participant")

Return both dictionaries as R objects:

dict <- data_dictionary("all")

Participant summaries are derived from the packaged 1-hour epoch data. For each sleep/wake method, a sleep/non-wear cutoff is selected from 50% to 100% in 0.01-point increments by matching estimated sleep/non-wear hours to weighted self-reported sleep. The selected cutoff is applied at the 1-hour epoch, and daily sedentary, waking, and sleep/non-wear hours are averaged across each participant's valid days.

Within-participant SD and SE columns are missing for participants with only one valid day because day-to-day variability cannot be estimated from a single observed day.

If you use wristsed, please cite the package, the accompanying manuscript, and the source methods for posture and sleep/non-wear classification.

In R, run:

citation("wristsed")

The package data product accompanies the manuscript:

Odoom C. Sedentary Behavior in the United States Measured by Accelerometer, NHANES 2011--2014. Manuscript in preparation. 2026.

The derived datasets build on NHANES wrist accelerometer data, CHAP posture classification methods, and SWaN sleep/wake/non-wear classification. Full citation details are available through citation("wristsed").

Relevant method DOIs:

• CHAP-Adult: https://doi.org/10.1123/jmpb.2021-0062
• CHAP validation study: https://doi.org/10.1249/MSS.0000000000002705
• SWaN sleep/non-wear method: https://doi.org/10.1249/MSS.0000000000002973

This package is licensed under the Apache License 2.0. See the LICENSE file for details.