Skip to contents

Simulate longitudinal phenotype data

Source: R/simulate_longit_dataset.R
simulate_long_pheno_data.Rd

Generates a long-format phenotype data.frame for one or more cohorts/sites across multiple sessions / time points.

Usage

simulate_long_pheno_data(
  data_structure = list(cohort1 = list(sessions = c("01", "02"), n_subjects = 100),
    cohort2 = list(sessions = c("01", "02"), n_subjects = 150)),
  baseline = list(age = c(mean = 10, sd = 0.5), sex = c(levels = c("Male", "Female")),
    wisdom = c(mean = 0, sd = 1)),
  change = list(age = c(mean = 4, sd = 0.5), wisdom = c(mean = 1, sd = 0.5)),
  dropout = NULL,
  seed = 3108,
  verbose = TRUE
)

Arguments

data_structure

Named list specifying site/cohort structure. Each element must be a list with:

sessions

Character vector of unique session labels.

n_subjects

Positive integer, number of subjects.

baseline

Named list defining baseline distributions. Supports continuous covariates (c(mean, sd)) and categorical covariates (c(levels = ...)). See section Supported covariates.

change

Named list of c(mean, sd) specifying the per-wave mean shift and noise SD for longitudinal covariates. Only continuous variables should appear here.

dropout

Percent cumulative attrition by the final wave (Default: NULL = no attrition)

seed

Integer random seed for reproducibility.

verbose

Logical

Value

A data.frame in long format with one row per subject × session. Always contains columns site, id, time, folder_id, + all covariates declared in baseline.

Supported covariates

Covariates are declared through the baseline argument. The type of covariate is inferred from the names of each list element:

Continuous (e.g. age, wisdom)

Specify c(mean = m, sd = s). Baseline values are drawn from \(N(\var{m}, \var{s}^2)\).

Categorical (e.g. sex)

Specify c(levels = c("Male", "Female")). Each subject is assigned a level with equal probability and this value is held constant across sessions.

Change model

Follow-up values for continuous variables listed in change are computed as: $$y_{i,s} = y_{i,1} + (s-1)\bar{\delta} + \varepsilon, \quad \varepsilon \sim N(0, \sigma_\delta^2)$$ where \(s\) is the 1-based session index. The deviation from baseline is drawn independently at each session. The variance \(\sigma_\delta^2\) is constant across sessions; only the mean shift accumulates linearly.

See also

simulate_freesurfer_data(), simulate_longit_dataset()

Examples

pheno <- simulate_long_pheno_data(
  data_structure = list(
    GENR = list(sessions = c("01", "02"), n_subjects = 50)
  ),
  baseline = list(age = c(mean = 10, sd = 1), sex = c(levels = c("Male", "Female"))),
  change   = list(age = c(mean = 4, sd = 0.5))
)
#> ⠙ Generate phenotype data
#> ℹ 100 total observations (from 1 site and 2 waves)
#> ⠙ Generate phenotype data

#>   Variables: id, site, time, age, sex, and folder_id
#> ⠙ Generate phenotype data

#> ✔ Generate phenotype data [44ms]
#> 
head(pheno)
#>   id site time       age    sex         folder_id
#> 1  1 GENR   01 11.565062 Female GENR/sub-1_ses-01
#> 2  2 GENR   01  7.527207   Male GENR/sub-2_ses-01
#> 3  3 GENR   01 10.194473 Female GENR/sub-3_ses-01
#> 4  4 GENR   01 11.811407   Male GENR/sub-4_ses-01
#> 5  5 GENR   01 10.131829   Male GENR/sub-5_ses-01
#> 6  6 GENR   01  9.036134 Female GENR/sub-6_ses-01