## Article 1 outcome group definition script
## To be enriched from Statistics Denmark
## 
## Based on the ItMLiHSmar2022 course

library(Hmisc)
library(dplyr)
library(daDoctoR)
library(tidyselect)

# Setting final primary output from "pout"
if (pout=="drop"){
  X_tbl <- X_tbl|>
    mutate(group=pase_drop_fac)
  
  # print(quantile(as.numeric(X_tbl$pase_0)))
  # print(quantile(as.numeric(X_tbl$pase_6)))
  # print(summary(X_tbl$pase_0_cut))
  
  X_tbl_f <- X_tbl|>
    filter(pase_0_cut!=1)|>
    select(-starts_with("pase_"))
}

if (pout=="hop"){
  X_tbl <- X_tbl|>
    mutate(group=pase_hop_fac)
  
  # print(quantile(as.numeric(X_tbl$pase_0)))
  # print(quantile(as.numeric(X_tbl$pase_6)))
  # print(summary(X_tbl$pase_0_cut))
  
  X_tbl_f <- X_tbl|>
    filter(pase_6_cut!=1)|>
    select(-starts_with("pase_"))
}

# Dropping non-complete for analysis
Xy <- X_tbl_f|> 
  na.omit()|> # Keeping only complete observations
  select(-c(tci) # Left out of model as no present in drop-group
  )|>
  mutate(mrs_0=factor(ifelse(mrs_0==1,1,2))) # Sets binary mRS 0 to include in glmnet, 0 or above

label(Xy) = as.list(var.labels[match(names(Xy), names(var.labels))])

X<-dplyr::select(Xy,-c(group, -starts_with("pase_")) # Exclude primary outcome
                 )
y<-Xy$group