Calculates a variety of physical activity variables based on triaxial minute-to-minute accelerometer count values for individual participants. Assumes first 1440 minutes are day 1, next 1440 are day 2, and so on. If final day has less than 1440 minutes, it is excluded. A data dictionary for the variables created is available here: https://github.com/vandomed/accelerometry/blob/master/process_tri_dictionary.csv.
process_tri(counts, steps = NULL, nci_methods = FALSE, start_day = 1,
start_date = NULL, id = NULL, brevity = 1, hourly_var = "cpm",
hourly_wearmin = 0, hourly_normalize = FALSE, valid_days = 1,
valid_wk_days = 0, valid_we_days = 0, int_axis = "vert",
int_cuts = c(100, 760, 2020, 5999), cpm_nci = FALSE,
days_distinct = FALSE, nonwear_axis = "vert", nonwear_window = 60,
nonwear_tol = 0, nonwear_tol_upper = 99, nonwear_nci = FALSE,
weartime_minimum = 600, weartime_maximum = 1440,
active_bout_length = 10, active_bout_tol = 0, mvpa_bout_tol_lower = 0,
vig_bout_tol_lower = 0, active_bout_nci = FALSE, sed_bout_tol = 0,
sed_bout_tol_maximum = int_cuts[2] - 1, artifact_axis = "vert",
artifact_thresh = 25000, artifact_action = 1, weekday_weekend = FALSE,
return_form = "daily")Integer matrix with three columns of count values, e.g. vertical-axis counts, anteroposterior (AP)-axis counts, and mediolateral (ML)-axis counts.
Integer vector with steps.
Logical value for whether to set all arguments so as to replicate the data processing methods used in the NCI's SAS programs. More specifically:
valid_days = 4
valid_wk_days = 0
valid_we_days = 0
int_axis = "vert"
int_cuts = c(100, 760, 2020, 5999)
cpm_nci = TRUE
days_distinct = TRUE
nonwear_axis = "vert"
nonwear_window = 60
nonwear_tol = 2
nonwear_tolupper = 100
nonwear_nci = TRUE
weartime_minimum = 600
weartime_maximum = 1440
active_bout_length = 10
active_bout_tol = 2
mvpa_bout_tol_lower = 0
vig_bout_tol_lower = 0
active_bout_nci = TRUE
sed_bout_tol = 0
sed_bout_tol_maximum = 759
artifact_thresh = 32767
artifact_action = 3
If TRUE, you can still specify non-default values for brevity
and weekday_weekend.
Integer value specifying day of week for first day of monitoring, with 1 = Sunday, ..., 7 = Satuday.
Date for first day of monitoring, which function can use to
figure out start_day.
Numeric value specifying ID number of participant.
Integer value controlling the number of physical activity variables generated. Choices are 1 for basic indicators of physical activity volume, 2 for addditional indicators of activity intensities, activity bouts, sedentary behavior, and peak activity, and 3 for additional hourly count averages.
Character string specifying what hourly activity variable
to record, if brevity = 3. Choices are "counts_vert", "counts_ap",
"counts_ml", "counts_sum", "counts_vm", "cpm_vert", "cpm_ap", "cpm_ml",
"sed_min", "sed_bouted_10min", and "sed_breaks".
Integer value specifying minimum number of wear time minutes needed during a given hour to record a value for the hourly activity variable.
Logical value for whether to normalize hourly activity by number of wear time minutes.
Integer value specifying minimum number of valid days to be considered valid for analysis.
Integer value specifying minimum number of valid weekdays to be considered valid for analysis.
Integer value specifying minimum number of valid weekend days to be considered valid for analysis.
Character string specifying which axis should be used to classify intensities. Choices are "vert", "ap", "ml", "sum" (for triaxial sum), and "vm (for triaxial vector magnitude).
Numeric vector with four cutpoints from which five intensity
ranges are derived. For example, int_cuts = c(100, 760, 2020, 5999)
creates: 0-99 = intensity 1; 100-759 = intensity level 2; 760-2019 =
intensity 3; 2020-5998 = intensity 4; >= 5999 = intensity 5. Intensities 1-5
are typically viewed as sedentary, light, lifestyle, moderate, and vigorous.
Logical value for whether to calculate average counts per
minute by dividing average daily counts by average daily wear time, as
opposed to taking the average of each day's counts per minute value. Strongly
recommend leave as FALSE unless you wish to replicate the NCI's SAS
programs.
Logical value for whether to treat each day of data as distinct, as opposed to analyzing the entire monitoring period as one continuous segment.
Character string specifying which axis should be used to classify non-wear time. Choices are "vert", "ap", "ml", "sum" (for triaxial sum), and "vm" (for triaxial vector magnitude).
Integer value specifying minimum length of a non-wear period.
Integer value specifying tolerance for non-wear algorithm, i.e. number of minutes with non-zero counts allowed during a non-wear interval.
Integer value specifying maximum count value for a minute with non-zero counts during a non-wear interval.
Logical value for whether to use non-wear algorithm from NCI's SAS programs.
Integer value specifying minimum number of wear time minutes for a day to be considered valid.
Integer value specifying maximum number of wear time minutes for a day to be considered valid. The default is 1440, but you may want to use a lower value (e.g. 1200) if participants were instructed to remove devices for sleeping, but often did not.
Integer value specifying minimum length of an active bout.
Integer value specifying number of minutes with counts
outside the required range to allow during an active bout. If non-zero and
active_bout_nci = FALSE, specifying non-zero values for
mvpa_bout_tol_lower and vig_bout_tol_lower is highly
recommended. Otherwise minutes immediately before and after an active bout
will tend to be classified as part of the bout.
Integer value specifying lower cut-off for count values outside of required intensity range for an MVPA bout.
Integer value specifying lower cut-off for count values outside of required intensity range for a vigorous bout.
Logical value for whether to use algorithm from the NCI's SAS programs for classifying active bouts.
Integer value specifying number of minutes with counts outside sedentary range to allow during a sedentary bout.
Integer value specifying upper cut-off for count values outside sedentary range during a sedentary bout.
Character string specifying which axis should be used to identify artifacts (impossibly high count values). Choices are "vert", "ap", "ml", "sum" (for triaxial sum), and "vm" (for triaxial vector magnitude).
Integer value specifying the smallest count value that should be considered an artifact.
Integer value controlling method of correcting artifacts. Choices are 1 to exclude days with one or more artifacts, 2 to lump artifacts into non-wear time, 3 to replace artifacts with the average of neighboring count values, and 4 to take no action.
Logical value for whether to calculate averages for weekdays and weekend days separately (in addition to all valid days).
Character string controlling how variables are returned. Choices are "daily" for per-day summaries, "averages" for averages across all valid days, and "both" for a list containing both.
Numeric matrix or list of two numeric matrices, depending on
return_form.
National Cancer Institute. Risk factor monitoring and methods: SAS programs for analyzing NHANES 2003-2004 accelerometer data. Available at: http://riskfactor.cancer.gov/tools/nhanes_pam. Accessed Aug. 19, 2018.
# NOT RUN {
# Note that the 'tridata' dataset contains 7 days of fake triaxial
# accelerometer data
# Process data using default parameters and request per-day variables
accel.days <- process_tri(
counts = tridata,
return_form = "daily"
)
# Repeat, but request averages across all valid days
accel.averages <- process_tri(
counts = tridata,
return_form = "averages"
)
# Create per-day summary again, but with many more variables
accel.days2 <- process_tri(
counts = tridata,
brevity = 2,
return_form = "daily"
)
names(accel.days2)
# }
Run the code above in your browser using DataLab