R/cleanBIOGENtrials.fun.R

In htx-r/RiskModelNMApredictions:

##########################################################################################

# 1. I chose the 'objective relpase assessment' (PARAMCD='OBJREL') as a type of the relapse
# However, it is the same as 'all realpse assessment'.
# 2. Create a dummy varaible from the AVAL varaible for the 1 year relapses (AVISIT='0-1 Year')
# 3. Create a dummy varaible from the AVAL varaible for the 2 years relapses (AVISIT='Overall 0-2 Years')
# 4. Check if the calculated dummy variables are correct.
# 5. Check if the IPD compatible with AD (form MS file)

# PS: AVAL counts the number of the relapses to each type of the assessment in PARAMCD.

#########################################################################################

cleanBIOGENtrials.fun=function(datapath){
  library(readxl)
  
  ## Read the IPD biogen data
  
  # path
  adslpath=paste(datapath,"/adsl.csv",sep="")
  adarrpath=paste(datapath,"/adarr.csv",sep="")
  adrelpath=paste(datapath,"/adrel.csv",sep="")
  
  # data
  adsl <- read.csv(adslpath)
  adarr <- read.csv(adarrpath)
  adrel <- read.csv(adrelpath)
  
  ## 1. Create the dataset with only the 'objective relpase assessment' (PARAMCD='OBJREL')
  adarr_OBJREL <- adarr[adarr$PARAMCD=='OBJREL',]
  
  
  ## 2. Create a dummy varaible (from AVAL) for the within 1 year relapses RELAPSE01Year
  adarr_OBJREL01year <- adarr_OBJREL[adarr_OBJREL$AVISIT=='0-1 Year',]
  RELAPSE01Year <-  ifelse(adarr_OBJREL01year$AVAL>0,1,0)
  
  ## 3. Create a dummy varaible (from AVAL) for the within 2 years relapses RELAPSE02Year
  
  #%%% All the patients have been checked within overall 0-2 Years except one study (only 0-1 year).
  # so it is not straight forward as 0-1 year where all studies have data.
  
  # 3.1 I extract the row for each subject who has 'Overall 0-2 Years' assessment (adarr_OBJREL_02year 'list of dataframes') then
  nsubjects <- length(unique(adarr$USUBJID))
  usubjectID <- unique(adarr$USUBJID)
  
  adarr_OBJREL_02year_ds <- sapply(1:nsubjects, function(i) subset(adarr_OBJREL,subset = adarr_OBJREL$USUBJID == as.character(usubjectID[i]) & adarr_OBJREL$AVISIT=='Overall 0-2 Years'),
                                   simplify = F)
  
  # 3.2 I check the nrow() of each data frame, if nrow==0 then there is no assessment for 'Overall 0-2 Years'
  # If not (nrow==1), then I check if the patient relapsed or not ('SUBJRR' differ from zero or not).
  
  RELAPSE02Year <- 1:nsubjects
  for (i in 1:nsubjects) {
    
    if(nrow(adarr_OBJREL_02year_ds[[i]]) == 0){
      RELAPSE02Year[i] <- NA
    }else{
      if(adarr_OBJREL_02year_ds[[i]]$AVAL>0){
        RELAPSE02Year[i] <- 1
        
      }else{
        RELAPSE02Year[i] <- 0
        
      }
    }
    
  }
  
  ## 3.3 Add the two dummy variables (RELAPSE01Year and RELAPSE02Year) to ADSL dataset now it is adsl01
  adsl01 <- adsl
  adsl01$RELAPSE1year <- RELAPSE01Year
  adsl01$RELAPSE2year <- RELAPSE02Year
  return(list(adsl01=adsl01,adarr_OBJREL=adarr_OBJREL))
  
}