Skip to contents

Generalized Stein-Fojo longitudinal model with subject-level covariates

Source: R/LongitudinalGSFCov.R
LongitudinalGSFCov-class.Rd

Separate covariate predictors model the log-normal distribution parameters for baseline, shrinkage, growth, and the logit-normal treatment fraction.

Formula intercept columns are removed because each predictor has a separate intercept. linear, proportional, exponential, and log-linear parametrizations have the same meaning as in LongitudinalSteinFojoCov(). The mu_phi predictor is the location of the normal distribution on the logit scale; phi itself is obtained by applying the inverse-logit transform.

Usage 

LongitudinalGSFCov(
  mu_b_formula = ~study,
  omega_b_formula = ~study,
  mu_s_formula = ~arm,
  omega_s_formula = ~arm,
  mu_g_formula = ~arm,
  omega_g_formula = ~arm,
  mu_phi_formula = ~arm,
  omega_phi_formula = ~arm,
  mu_b_parametrization = "linear",
  omega_b_parametrization = "log-linear",
  mu_s_parametrization = "linear",
  omega_s_parametrization = "log-linear",
  mu_g_parametrization = "linear",
  omega_g_parametrization = "log-linear",
  mu_phi_parametrization = "linear",
  omega_phi_parametrization = "log-linear",
  mu_b_intercept_prior = prior_normal(log(60), 1),
  mu_b_coefficients_prior = prior_normal(0, 1),
  omega_b_intercept_prior = prior_normal(log(0.2), 1),
  omega_b_coefficients_prior = prior_normal(0, 1),
  mu_s_intercept_prior = prior_normal(log(0.5), 1),
  mu_s_coefficients_prior = prior_normal(0, 1),
  omega_s_intercept_prior = prior_normal(log(0.2), 1),
  omega_s_coefficients_prior = prior_normal(0, 1),
  mu_g_intercept_prior = prior_normal(log(0.3), 1),
  mu_g_coefficients_prior = prior_normal(0, 1),
  omega_g_intercept_prior = prior_normal(log(0.2), 1),
  omega_g_coefficients_prior = prior_normal(0, 1),
  mu_phi_intercept_prior = prior_normal(qlogis(0.5), 1),
  mu_phi_coefficients_prior = prior_normal(0, 1),
  omega_phi_intercept_prior = prior_normal(log(0.2), 1),
  omega_phi_coefficients_prior = prior_normal(0, 1),
  sigma = prior_lognormal(log(0.1), 1),
  scaled_variance = FALSE,
  centred_baseline = TRUE,
  centred_shrinkage = FALSE,
  centred_growth = FALSE,
  centred_phi = FALSE
)

Arguments

mu_b_formula

(formula) One-sided formula for subject-level covariate effects on the baseline mean.

omega_b_formula

(formula) One-sided formula for subject-level covariate effects on the baseline standard deviation.

mu_s_formula

(formula) One-sided formula for subject-level covariate effects on the shrinkage mean.

omega_s_formula

(formula) One-sided formula for subject-level covariate effects on the shrinkage standard deviation.

mu_g_formula

(formula) One-sided formula for subject-level covariate effects on the growth mean.

omega_g_formula

(formula) One-sided formula for subject-level covariate effects on the growth standard deviation.

mu_phi_formula

(formula) One-sided formula for subject-level covariate effects on the mean of the treatment fraction on the logit scale.

omega_phi_formula

(formula) One-sided formula for subject-level covariate effects on the standard deviation of the treatment fraction on the logit scale.

mu_b_parametrization

(character) Parametrization of the baseline mean predictor.

omega_b_parametrization

(character) Parametrization of the baseline standard deviation predictor. Only "exponential" and "log-linear" are supported.

mu_s_parametrization

(character) Parametrization of the shrinkage mean predictor.

omega_s_parametrization

(character) Parametrization of the shrinkage standard deviation predictor. Only "exponential" and "log-linear" are supported.

mu_g_parametrization

(character) Parametrization of the growth mean predictor.

omega_g_parametrization

(character) Parametrization of the growth standard deviation predictor. Only "exponential" and "log-linear" are supported.

mu_phi_parametrization

(character) Parametrization of the mean predictor for the treatment fraction on the logit scale.

omega_phi_parametrization

(character) Parametrization of the standard deviation predictor for the treatment fraction on the logit scale. Only "exponential" and "log-linear" are supported.

mu_b_intercept_prior

(Prior) Prior for the intercept of the baseline mean predictor.

mu_b_coefficients_prior

(Prior) Prior for the covariate coefficients of the baseline mean predictor.

omega_b_intercept_prior

(Prior) Prior for the intercept of the baseline standard deviation predictor.

omega_b_coefficients_prior

(Prior) Prior for the covariate coefficients of the baseline standard deviation predictor.

mu_s_intercept_prior

(Prior) Prior for the intercept of the shrinkage mean predictor.

mu_s_coefficients_prior

(Prior) Prior for the covariate coefficients of the shrinkage mean predictor.

omega_s_intercept_prior

(Prior) Prior for the intercept of the shrinkage standard deviation predictor.

omega_s_coefficients_prior

(Prior) Prior for the covariate coefficients of the shrinkage standard deviation predictor.

mu_g_intercept_prior

(Prior) Prior for the intercept of the growth mean predictor.

mu_g_coefficients_prior

(Prior) Prior for the covariate coefficients of the growth mean predictor.

omega_g_intercept_prior

(Prior) Prior for the intercept of the growth standard deviation predictor.

omega_g_coefficients_prior

(Prior) Prior for the covariate coefficients of the growth standard deviation predictor.

mu_phi_intercept_prior

(Prior) Prior for the intercept of the mean predictor for the treatment fraction on the logit scale.

mu_phi_coefficients_prior

(Prior) Prior for the covariate coefficients of the mean predictor for the treatment fraction on the logit scale.

omega_phi_intercept_prior

(Prior) Prior for the intercept of the standard deviation predictor for the treatment fraction on the logit scale.

omega_phi_coefficients_prior

(Prior) Prior for the covariate coefficients of the standard deviation predictor for the treatment fraction on the logit scale.

sigma

(Prior) Observation-error standard deviation prior.

scaled_variance

(logical) Whether to use multiplicative observation error.

centred_baseline

(logical) Whether to use centred parameterizations.

centred_shrinkage

(logical) Whether to use centred parameterizations.

centred_growth

(logical) Whether to use centred parameterizations.

centred_phi

(logical) Whether to use centred parameterizations.

Value

A LongitudinalGSFCov object.