Simulate a synthetic correlate-of-protection cohort — mbmm_simulate_cop

Generates a small synthetic dataset in the spirit of the simulation study in Sec. 3.2 of Do, McCarthy & Rogers: each subject has a baseline (standardized) antibody value; a latent productive-infection state Z is drawn from an antibody-dependent logistic gate; subjects with Z = 1 get a viral-load trajectory from the eclipse-phase model (mbmm_example_eclipse_simulate) under subject-specific true kinetic parameters, observed with log-normal noise and a limit of detection; subjects with Z = 0 get a flat near-LOD trajectory with an occasional single-point transient/abortive "blip" (mimicking a non-productive PCR-detectable event). A subset of the Z = 1 subjects is given a denser visit schedule and flagged as bank-eligible, mirroring the donor-richness threshold used for real reference banks (Sec. 4.3).

Usage

mbmm_simulate_cop_demo(
  n_subjects = 90,
  prevalence = 0.35,
  beta_true = -1.2,
  n_bank_donors = 18,
  sigma_obs = 0.35,
  lod = 3,
  seed = 1
)

Arguments

n_subjects: total number of subjects to simulate
prevalence: marginal probability that a "typical" (zAb = 0) subject is truly infected (Z = 1); combined with beta_true to set the gate intercept
beta_true: true log-odds effect of standardized baseline antibody on infection probability (negative = protective)
n_bank_donors: number of Z = 1 subjects given a dense visit schedule and flagged bank_eligible = TRUE
sigma_obs: log10-scale observation noise SD
lod: limit of detection (natural scale, e.g. copies/mL)
seed: RNG seed

Value

a list with elements obs_list (named list of per-subject data.frames with columns time, y, censored, lod), covariates (data.frame with id and baseline_antibody), and truth (data.frame with the true generating values, for checking recovery – never used by the fitting functions themselves)

Details

This function only *generates data*; it does not call any MBMM fitting functions. See vignette("mbmm-intro") or system.file("examples/01_simulated_correlate_of_protection.R", package = "mbmm") for the full analysis pipeline built on top of it.