Extract to function code to create AgD from IPD

[n8thangreen]

covariate_defns_ald <- list(
  PF_cont_1 = list(type = continuous(mean = meanX_BC[1], sd = sdX[1]),
                    role = "prognostic"),
  PF_cont_2 = list(type = continuous(mean = meanX_BC[2], sd = sdX[2]),
                    role = "prognostic"),
  EM_cont_1 = list(type = continuous(mean = meanX_EM_BC[1], sd = sdX_EM[1]),
                    role = "effect_modifier"),
  EM_cont_2 = list(type = continuous(mean = meanX_EM_BC[2], sd = sdX_EM[2]),
                    role = "effect_modifier")
)

BC.IPD <- simcovariates::gen_data(
  N = N,
  b_0 = b_0,
  b_trt = b_trt,
  covariate_defns = covariate_defns_ald,
  b_prognostic = b_prognostic,
  b_effect_modifier = b_effect_modifier,
  cor_matrix = cor_matrix,
  trt_assignment = list(prob_trt1 = allocation),
  family = binomial("logit"))

BC.IPD$trt <- factor(BC.IPD$trt, labels = c("C", "B"))

# covariate summary statistics
# assume same between treatments
cov.X <- 
  BC.IPD %>%
  as.data.frame() |> 
  dplyr::select(matches("^(PF|EM)"), trt) |> 
  tidyr::pivot_longer(
    cols = starts_with("PF") | starts_with("EM"),
    names_to = "variable",
    values_to = "value") |>
  group_by(variable) %>%
  summarise(
    mean = mean(value),
    sd = sd(value)
  ) %>%
  tidyr::pivot_longer(
    cols = c("mean", "sd"),
    names_to = "statistic",
    values_to = "value") %>%
  ungroup() |> 
  mutate(trt = NA)

# outcome
summary.y <- 
  BC.IPD |> 
  as.data.frame() |> 
  dplyr::select(y, trt) %>%
  tidyr::pivot_longer(cols = "y",
               names_to = "variable",
               values_to = "value") %>%
  group_by(variable, trt) %>%
  summarise(
    mean = mean(value),
    sd = sd(value),
    sum = sum(value)
  ) %>%
  tidyr::pivot_longer(
    cols = c("mean", "sd", "sum"),
    names_to = "statistic",
    values_to = "value") %>%
  ungroup()

# sample sizes
summary.N <- 
  BC.IPD |> 
  group_by(trt) |> 
  count(name = "N") |> 
  tidyr::pivot_longer(
    cols = "N",
    names_to = "statistic",
    values_to = "value") |> 
  mutate(variable = NA_character_) |> 
  dplyr::select(variable, statistic, value, trt)
  
ald_trial <- rbind.data.frame(cov.X, summary.y, summary.N)