R/generics.R, R/sampling.R
sample_controls.RdThis method is used on trial_sequence objects to read, subset and sample expanded data.
sample_controls(
object,
p_control,
period = NULL,
subset_condition = NULL,
seed = NULL
)
# S4 method for class 'trial_sequence'
sample_controls(
object,
p_control,
period = NULL,
subset_condition = NULL,
seed = NULL
)An object of class trial_sequence.
Probability of selecting a control.
An integerish vector of non-zero length to select trial period(s) or NULL (default) to
select all trial periods.
A string or NULL (default). subset_condition will be translated to a call
(in case the expanded data is saved as a data.table or in the csv format) or to a SQL-query
(in case the expanded data is saved as a duckdb file).
The operators "==", "!=", ">", ">=", "<", "<=", %in%", "&", "|" are supported.
Numeric vectors can be written as c(1, 2, 3) or 1:3. Variables are not supported.
Note: Make sure numeric vectors written as 1:3 are surrounded by spaces, e.g. a %in% c( 1:4 , 6:9 ),
otherwise the code will fail.
An integer seed or NULL (default).
Note: The same seed will return a different result depending on the class of the te_datastore object contained in the trial_sequence object.
An updated trial_sequence object, the data is stored in slot @outcome_data
as a te_outcome_data object.
# create a trial_sequence-class object
trial_itt_dir <- file.path(tempdir(), "trial_itt")
dir.create(trial_itt_dir)
trial_itt <- trial_sequence(estimand = "ITT") |>
set_data(data = data_censored) |>
set_outcome_model(adjustment_terms = ~ x1 + x2)
trial_itt_csv <- set_expansion_options(
trial_itt,
output = save_to_csv(file.path(trial_itt_dir, "trial_csvs")),
chunk_size = 500
) |>
expand_trials()
# sample_controls default behaviour returns all trial_periods
sample_controls(trial_itt_csv, p_control = 0.01)
#> Trial Sequence Object
#> Estimand: Intention-to-treat
#> Data
#> N: 725 observations from 89 patients
#> Key: <id>
#> id period treatment x1 x2 x3 x4 age age_s
#> <int> <int> <num> <num> <num> <int> <num> <num> <num>
#> 1: 1 0 1 1 1.146148362 0 0.7342030 36 0.08333333
#> 2: 1 1 1 1 0.002200337 0 0.7342030 37 0.16666667
#> ---
#> 724: 99 6 1 1 -0.033762356 1 0.5752681 71 3.00000000
#> 725: 99 7 0 0 -1.340496520 1 0.5752681 72 3.08333333
#> outcome censored eligible time_of_event first am_1 cumA switch
#> <num> <int> <num> <num> <lgcl> <num> <num> <num>
#> 1: 0 0 1 9999 TRUE 0 1 0
#> 2: 0 0 0 9999 FALSE 1 2 0
#> ---
#> 724: 0 0 0 7 FALSE 1 4 0
#> 725: 1 0 0 7 FALSE 1 4 1
#> regime_start time_on_regime eligible0 eligible1 wt
#> <int> <num> <num> <num> <num>
#> 1: 0 0 1 0 1
#> 2: 0 1 0 1 1
#> ---
#> 724: 5 1 0 1 1
#> 725: 7 2 0 1 1
#>
#> IPW for informative censoring:
#> - No weight model specified
#>
#> Expansion:
#> Chunk size: 500
#> Censor at switch: FALSE
#> First period: 0 | Last period: Inf
#>
#> A TE Datastore CSV object
#> N: 1558 observations
#> Periods: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
#> Path: /tmp/RtmpGCze7e/trial_itt/trial_csvs
#> Columns: id, trial_period, followup_time, outcome, weight, treatment, x1, x2, assigned_treatment
#>
#> Outcome model:
#> TE Outcome Model Object
#> Formula: outcome ~ assigned_treatment + x1 + x2 + followup_time + I(followup_time^2) + trial_period + I(trial_period^2)
#> Treatment_var: assigned_treatment
#> Adjustment_vars: x1 x2
#>
#> Use fit_msm() to fit the outcome model
#>
#> Outcome data
#> N: 28 observations from 21 patients in 6 trial periods
#> Periods: 0 1 2 3 5 10
#> Sampling control observations with probability: 0.01
#> id trial_period followup_time outcome weight treatment x1 x2
#> <int> <int> <int> <int> <int> <int> <int> <num>
#> 1: 15 0 0 1 1 1 0 -0.7365256
#> 2: 32 0 0 1 1 1 1 1.9861380
#> ---
#> 27: 74 5 13 0 1 0 0 0.2366130
#> 28: 54 10 1 0 1 0 1 -0.3501371
#> assigned_treatment sample_weight
#> <int> <num>
#> 1: 1 1
#> 2: 1 1
#> ---
#> 27: 0 100
#> 28: 0 100
# sample_controls can subset the data before sampling
sample_controls(
trial_itt_csv,
p_control = 0.2,
period = 1:10,
subset_condition = "followup_time %in% 1:20 & x2 < 1",
)
#> Trial Sequence Object
#> Estimand: Intention-to-treat
#> Data
#> N: 725 observations from 89 patients
#> Key: <id>
#> id period treatment x1 x2 x3 x4 age age_s
#> <int> <int> <num> <num> <num> <int> <num> <num> <num>
#> 1: 1 0 1 1 1.146148362 0 0.7342030 36 0.08333333
#> 2: 1 1 1 1 0.002200337 0 0.7342030 37 0.16666667
#> ---
#> 724: 99 6 1 1 -0.033762356 1 0.5752681 71 3.00000000
#> 725: 99 7 0 0 -1.340496520 1 0.5752681 72 3.08333333
#> outcome censored eligible time_of_event first am_1 cumA switch
#> <num> <int> <num> <num> <lgcl> <num> <num> <num>
#> 1: 0 0 1 9999 TRUE 0 1 0
#> 2: 0 0 0 9999 FALSE 1 2 0
#> ---
#> 724: 0 0 0 7 FALSE 1 4 0
#> 725: 1 0 0 7 FALSE 1 4 1
#> regime_start time_on_regime eligible0 eligible1 wt
#> <int> <num> <num> <num> <num>
#> 1: 0 0 1 0 1
#> 2: 0 1 0 1 1
#> ---
#> 724: 5 1 0 1 1
#> 725: 7 2 0 1 1
#>
#> IPW for informative censoring:
#> - No weight model specified
#>
#> Expansion:
#> Chunk size: 500
#> Censor at switch: FALSE
#> First period: 0 | Last period: Inf
#>
#> A TE Datastore CSV object
#> N: 1558 observations
#> Periods: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
#> Path: /tmp/RtmpGCze7e/trial_itt/trial_csvs
#> Columns: id, trial_period, followup_time, outcome, weight, treatment, x1, x2, assigned_treatment
#>
#> Outcome model:
#> TE Outcome Model Object
#> Formula: outcome ~ assigned_treatment + x1 + x2 + followup_time + I(followup_time^2) + trial_period + I(trial_period^2)
#> Treatment_var: assigned_treatment
#> Adjustment_vars: x1 x2
#>
#> Use fit_msm() to fit the outcome model
#>
#> Outcome data
#> N: 123 observations from 24 patients in 10 trial periods
#> Periods: 1 2 3 4 5 6 7 8 9 10
#> Subset condition: followup_time %in% 1:20 & x2 < 1
#> Sampling control observations with probability: 0.2
#> id trial_period followup_time outcome weight treatment x1 x2
#> <int> <int> <int> <int> <int> <int> <int> <num>
#> 1: 50 1 1 0 1 1 1 -0.3846825
#> 2: 54 1 1 0 1 0 0 -1.4326303
#> ---
#> 122: 54 9 10 0 1 0 1 -0.7796190
#> 123: 54 10 6 0 1 0 1 -0.3501371
#> assigned_treatment sample_weight
#> <int> <num>
#> 1: 1 5
#> 2: 0 5
#> ---
#> 122: 0 5
#> 123: 0 5
# delete after use
unlink(trial_itt_dir, recursive = TRUE)