, Cole Beck [aut]
, Zhouwen Liu [aut],
Ryan Moore [aut], Charles E. Matthews [aut], Maciej S.
Buchowski [aut]| Title: | Process Accelerometer Data for Physical Activity Measurement |
|---|---|
| Description: | It provides a function "wearingMarking" for classification of monitor wear and nonwear time intervals in accelerometer data collected to assess physical activity. The package also contains functions for making plot for accelerometer data and obtaining the summary of various information including daily monitor wear time and the mean monitor wear time during valid days. "deliveryPred" and "markDelivery" can classify days for ActiGraph delivery by mail; "deliveryPreprocess" can process accelerometry data for analysis by zeropadding incomplete days and removing low activity days; "markPAI" can categorize physical activity intensity level based on user-defined cut-points of accelerometer counts. It also supports importing ActiGraph AGD files with "readActigraph" and "queryActigraph" functions. |
| Authors: | Leena Choi [aut, cre]
, Cole Beck [aut]
, Zhouwen Liu [aut],
Ryan Moore [aut], Charles E. Matthews [aut], Maciej S.
Buchowski [aut] |
| Maintainer: | Leena Choi <[email protected]> |
| License: | GPL (>= 3) |
| Version: | 0.2-4 |
| Built: | 2024-12-11 07:20:59 UTC |
| Source: | CRAN |
It provides a function wearingMarking for classification of monitor
wear and nonwear time intervals in accelerometer data collected to assess
physical activity. The package also contains functions for making plot for
accelerometer data and obtaining the summary of various information including
daily monitor wear time and the mean monitor wear time during valid days.
The revised package version 0.2-2 improved the functions in the previous version
regarding speed and robustness. In addition, several functions were added:
markDelivery can classify days for ActiGraph delivery by mail;
markPAI can categorize physical activity intensity level based on
user-defined cut-points of accelerometer counts. It also supports importing ActiGraph
AGD files with readActigraph and queryActigraph functions.
The package also better supports time zones and daylight saving.
Classify wear and nonwear time status for accelerometer data by
epoch-by-epoch basis by wearingMarking.
Classify mail delivery and non-delivery day status for accelerometer data
by markDelivery.
Three options are available for the package: pa.validCut=600,
pa.timeStamp='TimeStamp', and pa.cts='axis1'. When these
options are specified (as in markDelivery), the other
functions will automatically respect these values as defaults. For
instance, the count variable in data(dataSec) is "counts".
Running options(pa.cts='counts') allows the user to avoid specifying the
"cts" argument in wearingMarking. The options for validCut and
timeStamp are rarely changed.
Shiny app called Actigraph can be used to visualize accelerometer data and summarize the data. Please see https://github.com/couthcommander/PhysicalActivityShiny.
Leena Choi [email protected], Cole Beck [email protected], Zhouwen Liu [email protected], Charles E. Matthews [email protected], and Maciej S. Buchowski [email protected]
Maintainer: Leena Choi [email protected]
Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.
Choi L, Ward SC, Schnelle JF, Buchowski MS. Assessment of wear/nonwear time classification algorithms for triaxial accelerometer. Med Sci Sports Exerc. 2012 Oct;44(10):2009-16.
Choi L, Chen KY, Acra SA, Buchowski MS. Distributed lag and spline modeling for predicting energy expenditure from accelerometry in youth. J Appl Physiol. 2010 Feb;108(2):314-27.
data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) options(pa.cts = 'counts') # change cnt variable from "axis1" to "counts" data1m = wearingMarking(dataset = mydata1m, frame = 90) sumVct(data1m, id="sdata1m") plotData(data=data1m) summaryData(data=data1m, validCut=600, perMinuteCts=1, markingString = "w")data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) options(pa.cts = 'counts') # change cnt variable from "axis1" to "counts" data1m = wearingMarking(dataset = mydata1m, frame = 90) sumVct(data1m, id="sdata1m") plotData(data=data1m) summaryData(data=data1m, validCut=600, perMinuteCts=1, markingString = "w")
The function collapses counts in data collected with a short epoch to make a data set with a longer epoch. For example, this function collapses data with 1-sec epoch to 10-sec epoch or 1-min epoch data.
dataCollapser(dataset, TS, by, col, func = sum, ...)dataCollapser(dataset, TS, by, col, func = sum, ...)
dataset |
The source dataset, in dataframe format, that needs to be collapsed. |
TS |
The column name for timestamp. |
by |
Epoch in seconds for a collapsed dataset. For example, to collapse second data to minute data, set by = 60; to collapse 10-second data to minute data, set by = 60. |
col |
The column name(s) to collapse. If not provided, will default to all numeric columns. |
func |
A method for collapsing counts. The default is the summation of counts. |
... |
Argument settings that to be used by user-defined "func" setting. |
A collapsed data with user specified epoch.
Zhouwen Liu [email protected]
Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.
data(dataSec) ## collapse 1-sec epoch data to 10-sec epoch data mydata10s = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 10) ## collapse 1-sec epoch data to 1-min epoch data mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60)data(dataSec) ## collapse 1-sec epoch data to 10-sec epoch data mydata10s = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 10) ## collapse 1-sec epoch data to 1-min epoch data mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60)
Approximately 3 days of accelerometer data collected with 1-sec epoch in the
correct data format that can be used by wearingMarking to classify
wear and nonwear time.
data(dataSec)data(dataSec)
A data frame with 238140 observations on the following 2 required variables.
A character vector, timestamp of accelerometer measurements
A numeric vector, counts as accelerometer measurements
Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.
data(dataSec)data(dataSec)
Approximately 15 days of 3-axis accelerometer data collected with 1-minute epoch.
data(deliveryData)data(deliveryData)
A data frame with 20987 observations on the following variables.
A character vector, timestamp of accelerometer measurements
A numeric vector, counts from axis1 of 3-axis accelerometer
A numeric vector, counts from axis2 of 3-axis accelerometer
A numeric vector, counts from axis3 of 3-axis accelerometer
A numeric vector, the number of steps
A numeric vector, the vector magnitude calculated from counts of axis1, axis2 and axis3
Cole Beck [email protected], Leena Choi [email protected]
data(deliveryData)data(deliveryData)
The function extracts multiple statistical features relevant for classification of days as delivery or human wear. The extracted features are: mean, variance, maximum, absolute change, absolute energy, proportion of trial completed, 95th quantile, skewness, and kurtosis.
deliveryFeatures(df, ...)deliveryFeatures(df, ...)
df |
A dataframe. The source accelerometry dataset, in dataframe format. |
... |
not used at this time |
Function works for data consisting of one or multiple unique trials.
A dataframe is returned with a row for each unique day and a column for each feature.
The input dataframe should have the following columns: ‘TimeStamp’, ‘axis1’, ‘axis2’, ‘axis3’, ‘vm’, where ‘vm’ is the vector magnitude of axes 1, 2, and 3. Dataframe should also be formatted to 60 second epoch.
Ryan Moore [email protected], Cole Beck [email protected], and Leena Choi [email protected]
data(deliveryData) deliveryDataProcessed <- deliveryPreprocess(df = deliveryData) deliveryDataFeats <- deliveryFeatures(df = deliveryDataProcessed)data(deliveryData) deliveryDataProcessed <- deliveryPreprocess(df = deliveryData) deliveryDataFeats <- deliveryFeatures(df = deliveryDataProcessed)
The function is a wrapper function that performs preprocessing, feature extraction, and delivery day prediction of an accelerometry dataset. The prediction model can be selected from one of three models, a Random Forest, a logistic regression, and a convolutional neural network (default: Random Forest).
deliveryPred(df, model = c("RF", "NN", "GLM"))deliveryPred(df, model = c("RF", "NN", "GLM"))
df |
A dataframe. The source accelerometry dataset, in dataframe format. |
model |
A character. Indicates which prediction model to use. ‘RF’ is a Random Forest. ‘GLM’ is a logistic regression, and ‘NN’ is a convolutional neural network. |
Function works for data consisting of one or multiple unique trials.
A dataframe is returned with a predicted probability of each day being a delivery activity day.
The input dataframe should have the following columns: ‘TimeStamp’, ‘axis1’, ‘axis2’, ‘axis3’, ‘vm’, where ‘vm’ is the vector magnitude of axes 1, 2, and 3. Dataframe should also be formatted to 60 second epoch.
The function uses the default preprocessing criteria used in the development of the predictive models.
Ryan Moore [email protected], Cole Beck [email protected], and Leena Choi [email protected]
deliveryPreprocess, deliveryFeatures, deliveryPrediction
data(deliveryData) predictions <- deliveryPred(df = deliveryData, model = "GLM")data(deliveryData) predictions <- deliveryPred(df = deliveryData, model = "GLM")
The function predicts the probability of each day in an accelerometry dataset being caused from delivery activity instead of human activity. The prediction model can be selected from one of three models, a Random Forest, a logistic regression, and a convolutional neural network (default: Random Forest).
deliveryPrediction(df, feats, model = c("RF", "GLM", "NN"), ...)deliveryPrediction(df, feats, model = c("RF", "GLM", "NN"), ...)
df |
A dataframe. The source accelerometry dataset, in dataframe format. |
feats |
A dataframe. Features output from the |
model |
A character. Indicates which prediction model to use. ‘RF’ is a Random Forest. ‘GLM’ is a logistic regression, and ‘NN’ is a convolutional neural network. |
... |
not used at this time |
Function works for data consisting of one or multiple unique trials.
A dataframe is returned with a predicted probability of each day being a delivery activity day.
The input dataframe should have the following columns: ‘TimeStamp’, ‘axis1’, ‘axis2’, ‘axis3’, ‘vm’, where ‘vm’ is the vector magnitude of axes 1, 2, and 3. Dataframe should also be formatted to 60 second epoch.
Ryan Moore [email protected], Cole Beck [email protected], and Leena Choi [email protected]
deliveryFeatures, deliveryPred
data(deliveryData) deliveryDataProcessed <- deliveryPreprocess(df = deliveryData) deliveryDataFeats <- deliveryFeatures(df = deliveryDataProcessed) deliveryPrediction(deliveryDataProcessed, deliveryDataFeats)data(deliveryData) deliveryDataProcessed <- deliveryPreprocess(df = deliveryData) deliveryDataFeats <- deliveryFeatures(df = deliveryDataProcessed) deliveryPrediction(deliveryDataProcessed, deliveryDataFeats)
This function preprocesses accelerometry data by removing days based on a total activity count (default: less than 5000) or total time with activity (default: less than 10 minutes). Additionally, the function has an option to zeropad truncated days such that that days that do not have a whole day of 1440 minutes of data spanning from 00:00 to 23:59 (default: TRUE).
deliveryPreprocess(df, minLow = 5000, minTime = 10, zeropad = TRUE, ...)deliveryPreprocess(df, minLow = 5000, minTime = 10, zeropad = TRUE, ...)
df |
A dataframe. The source accelerometry dataset, in dataframe format. |
minLow |
Numeric. The minimum total counts of movement for a day to not be removed. |
minTime |
Numeric. The minimum number of minutes of activity for a day to not be removed. |
zeropad |
Boolean value for truncated days to be zeropadded. |
... |
not used at this time |
Function works for dataframes from one or multiple unique trials.
The dataframe is returned with days fulfilling the dropping criteria removed and truncated days zeropadded. A new column indicating which day is added to the dataframe.
The input dataframe should have the following columns: ‘TimeStamp’, ‘axis1’, ‘axis2’, ‘axis3’, ‘vm’, where ‘vm’ is the vector magnitude of axes 1, 2, and 3. Dataframe should also be formatted to 60 second epoch.
Ryan Moore [email protected], Cole Beck [email protected], and Leena Choi [email protected]
data(deliveryData) deliveryDataProcessed <- deliveryPreprocess(df = deliveryData)data(deliveryData) deliveryDataProcessed <- deliveryPreprocess(df = deliveryData)
This function adds an indicator variable for accelerometer delivery days based
on a delivery classification algorithm. The algorithm classifies each day as
delivery or non-delivery day within each participant data using summary
statistics of accelerometer counts for each day. As the summary statistics,
the 95th percentile, mean and standard deviation (sd) of accelerometer counts
can be used. Using the summary statistics for each day, the algorithm defines
a set of days that are used to estimate the 95% confidence interval (CI)
based on t-distribution (default) or normal distribution. The lower bound of
the 95% CI is used to classify delivery days; if the summary statistics
for a day is below the lower bound of the 95% CI, this day is classified
as delivery day. Three methods for defining a set of days are available:
trim (default), consecutive, and valid.
markDelivery( data, cts = getOption("pa.cts"), markingString = "w", window = c("trim", "consecutive", "valid"), method = c("95", "mean", "sd"), validCut = getOption("pa.validCut"), wearThreshold = 300, dist = c("t", "normal") )markDelivery( data, cts = getOption("pa.cts"), markingString = "w", window = c("trim", "consecutive", "valid"), method = c("95", "mean", "sd"), validCut = getOption("pa.validCut"), wearThreshold = 300, dist = c("t", "normal") )
data |
Data with classified wear (nonwear) status by
|
cts |
The name of the counts column. The default is “axis1”. |
markingString |
Option for summarizing wear (markingString = “w”) or nonwear time (markingString = “nw”). |
window |
A character. It should be one of ‘trim’, ‘consecutive’, or ‘valid’. |
method |
A character. It should be one of ‘95’, ‘mean’ or ‘sd’. |
validCut |
A cutoff for the total minutes of classified monitor wear time per day to be considered as a valid monitor day. |
wearThreshold |
A numeric value specifying a pseudo-valid day cutoff similar to “validCut”, which is used to define a set of days to estimate the 95% CI. |
dist |
Option for distribution used to calculate the 95% CI. |
A data frame with summary information about daily counts.
Cole Beck [email protected], Leena Choi [email protected]
data(deliveryData) options(pa.cts = "vm") wm <- wearingMarking(dataset = deliveryData) markDelivery(wm) plotData(data=wm) # days 1, 2, 10 - 15 are delivery or invalid days based on the result above markDelivery(wm, window='valid', method='mean') markDelivery(wm, method='mean') markDelivery(wm, method='sd')data(deliveryData) options(pa.cts = "vm") wm <- wearingMarking(dataset = deliveryData) markDelivery(wm) plotData(data=wm) # days 1, 2, 10 - 15 are delivery or invalid days based on the result above markDelivery(wm, window='valid', method='mean') markDelivery(wm, method='mean') markDelivery(wm, method='sd')
This function adds a "day" variable to the source dataset. The day is marked in numeric order, according to the timestamp variable.
markingTime(dataset, timestamp, startTime = "00:00:00", tz = "UTC")markingTime(dataset, timestamp, startTime = "00:00:00", tz = "UTC")
dataset |
The source dataset, in dataframe format, which needs to be marked. |
timestamp |
The column name in the dataset that will be used as timestamp. |
startTime |
Define the starting time of a day. It must be in the format of "hh:mm:ss". |
tz |
Local time zone, defaults to UTC. |
A dataframe with an extra day marking column.
Zhouwen Liu [email protected]
Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.
This function adds a physical activity intensity level variable “pai” to the source dataset. The “pai” is an ordered factor variable. It will be NA for nonwear times.
markPAI( data, cts = getOption("pa.cts"), markingString = "w", breaks = c(-Inf, 100, 760, 2020, Inf), labels = c("sedentary", "light", "moderate", "vigorous") )markPAI( data, cts = getOption("pa.cts"), markingString = "w", breaks = c(-Inf, 100, 760, 2020, Inf), labels = c("sedentary", "light", "moderate", "vigorous") )
data |
Data with classified wear (nonwear) status by
|
cts |
The name of the counts column. The default is “axis1”. |
markingString |
Option for summarizing wear (markingString = “w”) or nonwear time (markingString = “nw”). |
breaks |
A numeric vector of cut-points. The default cut-points are based on Matthews et al. (2016). |
labels |
A character vector labelling intensity levels. |
A data frame with an additional PAI-level column.
Cole Beck [email protected], Leena Choi [email protected]
Matthews CE, Keadle SK, Troiano RP, Kahle L, Koster A, Brychta R, Van Domelen D, Caserotti P, Chen KY, Harris TB, Berrigan D. Accelerometer-measured dose-response for physical activity, sedentary time, and mortality in US adults. Am J Clin Nutr. 2016 Nov;104(5):1424-1432.
data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, perMinuteCts = 1, cts = "counts") markPAI(data = data1m, cts = 'counts')[1:10,]data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, perMinuteCts = 1, cts = "counts") markPAI(data = data1m, cts = 'counts')[1:10,]
This function makes plot for accelerometer collected data (counts) over time for the whole monitor period, or a user specified time period or day with a midnight marking to separate monitored days.
plotData( data, day = NULL, start = NULL, end = NULL, cts = getOption("pa.cts"), TS = getOption("pa.timeStamp"), summary = NULL )plotData( data, day = NULL, start = NULL, end = NULL, cts = getOption("pa.cts"), TS = getOption("pa.timeStamp"), summary = NULL )
data |
Data with classified wear and nonwear status from
|
day |
A part of data during a user specified day for plot. |
start |
Define a starting time for plot. |
end |
Define a ending time for plot. |
cts |
The name of the counts column. The default is "axis1". |
TS |
The column name for timestamp. The default is "TimeStamp". |
summary |
List output of |
If a local time-zone is specified for wearkingMarking, it is possible
that daylight savings starts or ends during the period shown. In this case
a dotted line will indicate its position and the appropriate time-zone
abbreviations will be included.
Plot with midnight marking.
Leena Choi [email protected]
wearingMarking, sumVct, summaryData
data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing") ## change "cts" default from "axis1" to "counts" options(pa.cts = "counts") ## plot the whole data plotData(data=data1m) ## plot the data from 60 min to 900 min plotData(data=data1m, start=60, end=900) ## plot the data for day 2 plotData(data=data1m, day=2) ## include summaryData sumdat <- summaryData(data=data1m) plotData(data=data1m, summary=sumdat) ## present daylight saving time change data(deliveryData) options(pa.cts = "vm") wm <- wearingMarking(dataset = deliveryData, TS="TimeStamp", tz="America/Chicago") sumdat <- summaryData(wm) plotData(data=wm, summary = sumdat) ## valid data after delivery marking del <- markDelivery(wm) sumdat <- summaryData(wm, delivery = del) plotData(data=wm, summary = sumdat)data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing") ## change "cts" default from "axis1" to "counts" options(pa.cts = "counts") ## plot the whole data plotData(data=data1m) ## plot the data from 60 min to 900 min plotData(data=data1m, start=60, end=900) ## plot the data for day 2 plotData(data=data1m, day=2) ## include summaryData sumdat <- summaryData(data=data1m) plotData(data=data1m, summary=sumdat) ## present daylight saving time change data(deliveryData) options(pa.cts = "vm") wm <- wearingMarking(dataset = deliveryData, TS="TimeStamp", tz="America/Chicago") sumdat <- summaryData(wm) plotData(data=wm, summary = sumdat) ## valid data after delivery marking del <- markDelivery(wm) sumdat <- summaryData(wm, delivery = del) plotData(data=wm, summary = sumdat)
This function executes a SELECT query on an ActiGraph AGD file.
queryActigraph(datfile, qry)queryActigraph(datfile, qry)
datfile |
An AGD file. |
qry |
An SQL SELECT statement. |
AGD files are actually SQLite databases. This function requires the RSQLite package. The user is encouraged to directly interface with the database by creating a connection with the DBI package. This has been tested with AGD files produced with ActiLife v6.11.
A data frame with query results.
Cole Beck [email protected]
## Not run: dat <- queryActigraph("actfile.agd", "SELECT * FROM data LIMIT 5") queryActigraph("actfile.agd", "SELECT * FROM settings") ## directly interface using DBI package con <- DBI::dbConnect(RSQLite::SQLite(), "actfile.agd") DBI::dbListTables(con) DBI::dbDisconnect(con) ## End(Not run)## Not run: dat <- queryActigraph("actfile.agd", "SELECT * FROM data LIMIT 5") queryActigraph("actfile.agd", "SELECT * FROM settings") ## directly interface using DBI package con <- DBI::dbConnect(RSQLite::SQLite(), "actfile.agd") DBI::dbListTables(con) DBI::dbDisconnect(con) ## End(Not run)
This function reads an ActiGraph AGD file into R as a data frame.
If accelerometer data are collected with three axes, it creates vector magnitude
(vm). The counts at any axis or "vm" can be used to classify with wear and
nonwear time using wearingMarking.
readActigraph(datfile, convertTime = TRUE)readActigraph(datfile, convertTime = TRUE)
datfile |
An AGD file. |
convertTime |
Convert the timestamp from a character string into POSIXct. |
AGD files are SQLite databases. This function requires the RSQLite package.
A data frame with accelerometer data.
Cole Beck [email protected]
wearingMarking, queryActigraph
## Not run: dat <- readActigraph("actfile.agd") dat1s <- wearingMarking(dataset = dat, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "axis1", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing", getMinuteMarking = FALSE) ## End(Not run)## Not run: dat <- readActigraph("actfile.agd") dat1s <- wearingMarking(dataset = dat, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "axis1", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing", getMinuteMarking = FALSE) ## End(Not run)
This function converts accelerometer output data to a correct data format to
classify wear and nonwear time using wearingMarking. This
function can accept accelerometer output data with various epochs (for
example, 1-sec, 10-sec or 1-min). If accelerometer data are collected with
three axes, it creates vector magnitude (vm).
readCountsData(filename, ctPerSec, mode = 0)readCountsData(filename, ctPerSec, mode = 0)
filename |
A filename of accelerometer output to be read. |
ctPerSec |
Data collection epoch. This argument tells the program the number of counting will be performed in every second. For examples: for 1-sec epoch data, set ctPerSec = 1; for 10-sec epoch data, set ctPerSec = 1/10; for 1-min epoch data, set ctPerSec = 1/60. |
mode |
The mode of the ActiLife dat file. Defaults to 0, and should be listed in the file header. |
a data frame with the correct format (TimeStamp, counts) to be used for
wearingMarking.
Warning: It can be very slow if accelerometer data were collected with 1-sec epoch for many days.
Zhouwen Liu [email protected]
Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.
############################################################################### ## Read accelerometer output and convert to a correct format (TimeStamp, counts) ## Suppose "rawActigraphOutput.dat" is an Actigraph output with header as follows: ############################################################################### ## --- Data File Created By ActiGraph GT1M ActiLife v4.4.1 Firmware v7.2.0 --- ## Serial Number: LYN2B21080027 ## Start Time 16:15:00 ## Start Date 6/16/2010 ## Epoch Period (hh:mm:ss) 00:00:01 ## Download Time 09:50:23 ## Download Date 6/22/2010 ## Current Memory Address: 983038 ## Current Battery Voltage: 4.01 Mode = 0 ## -------------------------------------------------- ############################################################################### ## This raw data with 1-sec epoch can be converted to a correct data format to ## classify wear and nonwear time using "wearingMarking" by the following code: ## Not run: mydata1s = readCountsData("rawActigraphOutput.dat", ctPerSec=1)############################################################################### ## Read accelerometer output and convert to a correct format (TimeStamp, counts) ## Suppose "rawActigraphOutput.dat" is an Actigraph output with header as follows: ############################################################################### ## --- Data File Created By ActiGraph GT1M ActiLife v4.4.1 Firmware v7.2.0 --- ## Serial Number: LYN2B21080027 ## Start Time 16:15:00 ## Start Date 6/16/2010 ## Epoch Period (hh:mm:ss) 00:00:01 ## Download Time 09:50:23 ## Download Date 6/22/2010 ## Current Memory Address: 983038 ## Current Battery Voltage: 4.01 Mode = 0 ## -------------------------------------------------- ############################################################################### ## This raw data with 1-sec epoch can be converted to a correct data format to ## classify wear and nonwear time using "wearingMarking" by the following code: ## Not run: mydata1s = readCountsData("rawActigraphOutput.dat", ctPerSec=1)
This function summarizes accelerometer data and the classified wear or nonwear time by daily basis.
summaryData( data, validCut = getOption("pa.validCut"), perMinuteCts = 1, markingString = "w", TS = getOption("pa.timeStamp"), cts = getOption("pa.cts"), delivery = NULL, deliveryCut = 0.5 )summaryData( data, validCut = getOption("pa.validCut"), perMinuteCts = 1, markingString = "w", TS = getOption("pa.timeStamp"), cts = getOption("pa.cts"), delivery = NULL, deliveryCut = 0.5 )
data |
Data with classified wear (nonwear) status by
|
validCut |
A cutoff for the total minutes of classified monitor wear time per day to be considered as a valid monitor day. |
perMinuteCts |
The number of data rows per minute. The default is 1-min epoch (perMinuteCts = 1) and we recommend to use 1-min epoch data for this summary. For examples: for data with 10-sec epoch, set perMinuteCts = 6; for data with 1-sec epoch, set perMinuteCts = 60. |
markingString |
Option for summarizing wear (markingString = “w”) or nonwear time (markingString = “nw”). |
TS |
The column name for timestamp. The default is “TimeStamp”. |
cts |
The name of the counts column. The default is “axis1”. |
delivery |
data.frame. Delivery information created by
|
deliveryCut |
A cutoff (probability) to consider a valid delivery date.
See the |
This function summarizes the total number of days, weekdays and weekend days
in accelerometer data. It provides the total number of valid days,
valid weekdays and valid weekend days based on a user defined cutoff
for the total minutes of classified monitor wear time per day. This function
also summarizes the classified wear (nonwear) time by day and by valid day,
and the mean wear (nonwear) time for valid days during weekday and weekends,
and for overall valid days. If mail delivery days are classified by
markDelivery, it also summarizes the classified delivery
(non-delivery) days with argument “delivery”.
If “pai” column is present in the data,
which can be created by markPAI, then physical activity intensity
(PAI) level will be summarized in the output.
unit |
epoch for data. |
totalNumDays |
the total number of days in accelerometer data. |
totalNumWeekWeekend |
the total number of weekdays and weekend days in accelerometer data. |
validCut |
a user defined cutoff for the total minutes of classified monitor wear time per day to be considered as a valid monitor day. |
totalValidNumDays |
the total number of valid days based on the user defined cutoff (“validCut”) for the total minutes of wear time and the classified wear time. |
totalValidNumWeekWeekend |
the total number of valid weekdays and valid weekend days based on the user defined cutoff (“validCut”) for the total minutes of classified monitor wear time per day. |
wearTimeByDay |
the classified total wear (nonwear) time by day. |
deliveryDays |
marked delivery days. |
validWearTimeByDay |
the classified total wear (nonwear) time by valid day. |
meanWeartimeValidDays |
the mean wear (nonwear) time for valid days during weekdays and weekends. |
meanWeartimeOverallValidDays |
the mean wear (nonwear) time for overall valid days. |
dayInfo |
information about wear time and mean counts for each day. |
intensity |
optional output depending on “pai” column in the data; the total time in hours of physical activity intensity by day. |
meanValidIntensity |
optional output depending on “pai” column in the data; the mean physical activity intensity (PAI) level for valid days. |
Cole Beck [email protected], Leena Choi [email protected]
Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.
wearingMarking, sumVct,
markPAI, markDelivery
data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing") summaryData(data=data1m, validCut=600, perMinuteCts=1, markingString = "w", cts = "counts") data(deliveryData) options(pa.cts = "vm") wm <- wearingMarking(dataset = deliveryData) dd <- markDelivery(wm) pdd <- deliveryPred(wm) summaryData(wm, delivery = dd) summaryData(wm, delivery = pdd) pai.data <- markPAI(data = wm) dd <- markDelivery(pai.data) summaryData(pai.data, delivery = dd)data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing") summaryData(data=data1m, validCut=600, perMinuteCts=1, markingString = "w", cts = "counts") data(deliveryData) options(pa.cts = "vm") wm <- wearingMarking(dataset = deliveryData) dd <- markDelivery(wm) pdd <- deliveryPred(wm) summaryData(wm, delivery = dd) summaryData(wm, delivery = pdd) pai.data <- markPAI(data = wm) dd <- markDelivery(pai.data) summaryData(pai.data, delivery = dd)
This function summarizes the classified wear (nonwear) time by interval basis
from the epoch-by-epoch classified wear (nonwear) status classified by
wearingMarking.
sumVct( datavct, wearing = "wearing", TS = getOption("pa.timeStamp"), markingString = "w", by = "days", id = NULL )sumVct( datavct, wearing = "wearing", TS = getOption("pa.timeStamp"), markingString = "w", by = "days", id = NULL )
datavct |
Data with classified wear (nonwear) status classified by
|
wearing |
The column name for classified wear and nonwear status. The default is "wearing". |
TS |
The column name for timestamp. The default is "TimeStamp". |
markingString |
Option for summarizing wear (markingString="w") or nonwear time interval (markingString="nw"). |
by |
A sequence of days for classified wear (nonwear) time intervals. |
id |
Optional output for subject identification or file name. |
The summary data for wear and nonwear time intervals.
Leena Choi [email protected], Cole Beck [email protected], Zhouwen Liu [email protected], Charles E. Matthews [email protected], and Maciej S. Buchowski [email protected]
Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.
data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing") sumVct(data1m, id="sdata1m") sumVct(data1m, id="sdata1m", markingString = "nw")data(dataSec) mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing") sumVct(data1m, id="sdata1m") sumVct(data1m, id="sdata1m", markingString = "nw")
This function classifies wear and nonwear time status for accelerometer data by epoch-by-epoch basis.
wearingMarking( dataset, frame = 90, perMinuteCts = 60, TS = getOption("pa.timeStamp"), cts = getOption("pa.cts"), streamFrame = NULL, allowanceFrame = 2, newcolname = "wearing", getMinuteMarking = FALSE, dayStart = "00:00:00", tz = "UTC", ... )wearingMarking( dataset, frame = 90, perMinuteCts = 60, TS = getOption("pa.timeStamp"), cts = getOption("pa.cts"), streamFrame = NULL, allowanceFrame = 2, newcolname = "wearing", getMinuteMarking = FALSE, dayStart = "00:00:00", tz = "UTC", ... )
dataset |
The source dataset, in dataframe format, which needs to be marked. |
frame |
The size of time interval to be considered; Window 1 described in Choi et al. (2011). The default is 90. |
perMinuteCts |
The number of data rows per minute. The default is 1-sec epoch (perMinuteCts = 60). For examples: for data with 10-sec epoch, set perMinuteCts = 6; for data with 1-min epoch, set perMinuteCts = 1. |
TS |
The column name for timestamp. The default is “TimeStamp”. |
cts |
The column name for counts. The default is “axis1”. |
streamFrame |
The size of time interval that the program will look back or forward if activity is detected; Window 2 described in Choi et al. (2011). The default is the half of the frame. |
allowanceFrame |
The size of time interval that zero counts are allowed; the artifactual movement interval described in Choi et al. (2011). The default is 2. |
newcolname |
The column name for classified wear and nonwear status. The default is “wearing”. After the data is processed, a new field will be added to the original dataframe. This new field is an indicator for the wearing (“w”) or nowwearing (“nw”). |
getMinuteMarking |
Return minute data with wear and nonwear classification. If the source is not a minute dataset, the function will collapse it into minute data. The default is FALSE. |
dayStart |
Define the starting time of day. The default is the midnight, "00:00:00". It must be in the format of "hh:mm:ss". |
tz |
Local time zone, defaults to UTC. |
... |
Parameter settings that will be used in
|
A detailed description of the algorithm implemented in this function is described in Choi et al. (2011).
A data frame with the column for wear and nonwear classification indicator by epoch-by-epoch basis.
Warning: It will be very slow if accelerometer data with 1-sec epoch
for many days are directly classified. We recommend to collapse a dataset
with 1-sec epoch to 1-min epoch data using dataCollapser and
then classify wear and nonwear status using a dataset with a larger epoch.
Leena Choi [email protected], Cole Beck [email protected], Zhouwen Liu [email protected], Charles E. Matthews [email protected], and Maciej S. Buchowski [email protected]
Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.
data(dataSec) ## mark data with 1-min epoch mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing") sumVct(data1m, id="dataid") ## mark data with 1-sec epoch ## Not run: data1s = wearingMarking(dataset = dataSec, frame = 90, perMinuteCts = 60, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing", getMinuteMarking = FALSE) sumVct(data1s, id="dataid") sumVct(data1s, id="dataid", markingString = "nw") ## End(Not run)data(dataSec) ## mark data with 1-min epoch mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60) data1m = wearingMarking(dataset = mydata1m, frame = 90, perMinuteCts = 1, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing") sumVct(data1m, id="dataid") ## mark data with 1-sec epoch ## Not run: data1s = wearingMarking(dataset = dataSec, frame = 90, perMinuteCts = 60, TS = "TimeStamp", cts = "counts", streamFrame = NULL, allowanceFrame= 2, newcolname = "wearing", getMinuteMarking = FALSE) sumVct(data1s, id="dataid") sumVct(data1s, id="dataid", markingString = "nw") ## End(Not run)