cleanup_data {breathtestcore}

Transforms 13C breath data into a clean format for fitting

Description

Accepts various data formats of ungrouped or grouped 13C breath test time series, and transforms these into a data frame that can be used by all fitting functions, e.g. nls_fit. If in doubt, pass data frame through cleanup_data before forwarding it to a fitting function. If the function cannot repair the format, it gives better error messages than the xxx_fit functions.

Usage

cleanup_data(data, ...)

Arguments

data

• A data frame, array or tibble with at least two numeric columns with optional names minute and pdr to fit a single 13C record.

• A data frame or tibble with three columns named patient_id, minute and pdr.

• A matrix that can be converted to one of the above.

• A list of data frames/tibbles that are concatenated. When the list has named elements, the names are converted to group labels. When the list elements are not named, group name A is used for all items.

• A structure of class breathtest_data, as imported from a file with read_any_breathtest

• A list of class breathtest_data_list as generated from read function such as read_breathid_xml

...

optional.

use_filename_as_patient_id

Always use filename instead of patient name. Use this when patient id are not unique.

Value

A tibble with 4 columns. Column patient_id is created with a dummy entry of pat_a if no patient_id was present in the input data set. A column group is required in the input data if the patients are from different treatment groups or within-subject repeats, e.g. in crossover design. A dummy group name "A" is added if no group column was available in the input data set. If group is present, this is a hint to the analysis functions to do post-hoc breakdown or use it as a grouping variable in population-based methods. A patient can have records in multiple groups, for example in a cross-over designs.

Columns minute and pdr are the same as given on input, but negative minute values are removed, and an entry at 0 minutes is shifted to 0.01 minutes because most fit methods cannot handle the singularity at t=0.

An error is raised if dummy columns patient_id and group cannot be added in a unique way, i.e. when multiple values for a given minute cannot be disambiguated.

Comments are persistent; multiple comments are concatenated with newline separators.

Examples

options(digits = 4)
# Full manual
minute = seq(0,30, by = 10)
data1 = data.frame(minute, 
   pdr = exp_beta(minute, dose = 100, m = 30,  k = 0.01, beta = 2))
# Two columns with data at t = 0
data1
# Four columns with data at t = 0.01
cleanup_data(data1)

# Results from simulate_breathtest_data can be passed directly to cleanup_data
cleanup_data(simulate_breathtest_data(3))
# .. which implicitly does
cleanup_data(simulate_breathtest_data(3)$data)

# Use simulated data
data2 = list(
  Z = simulate_breathtest_data(seed = 10)$data,
  Y = simulate_breathtest_data(seed = 11)$data)
d = cleanup_data(data2)
str(d)
unique(d$patient_id)
unique(d$group)
# "Z" "Y"

# Mix multiple input formats
f1 = btcore_file("350_20043_0_GER.txt")
f2 = btcore_file("IrisMulti.TXT")
f3 = btcore_file("IrisCSV.TXT")
# With a named list, the name is used as a group parameter
data = list(A = read_breathid(f1), B = read_iris(f2), C = read_iris_csv(f3)) 
d = cleanup_data(data)
str(d)
unique(d$patient_id)
# "350_20043_0_GER" "1871960"         "123456"
# File name is used as patient name if none is available
unique(d$group)
# "A" "B" "C"