R/coxTimeToEvent.R

In csmpv: Biomarker Confirmation, Selection, Modelling, Prediction, and Validation

#' A function rather aimed at developers
#' @import survminer
#' @import survival
#' @import ggplot2
#' @import ggpubr
#' @importFrom grDevices dev.off
#' @importFrom grDevices pdf
#' @importFrom stats as.formula
#' @noRd

coxTimeToEvent = function(datain, time, event, Xin, outfile = "nameWithPath"){
  ## datain is a data frame
  ## the outcome type is time to event, so, there are two outcome variable names: time and event
  ## Xin is one x variable, or a group of x variables
  
  survY = paste0("survival::Surv(", time,",", event, ")")
  survX = paste(Xin, collapse=" + ")
  
  newplot = NA
  
  if(length(Xin) == 1){
    tmp = table(datain[,Xin])
    if(length(tmp) <= 4){
      fit = survminer::surv_fit(as.formula(paste(survY, survX, sep=" ~ ")), data = datain)
      newplot <- survminer::ggsurvplot(
        fit,
        risk.table = TRUE,
        pval = TRUE,
        conf.int = FALSE,
        palette = c("#2E9FDF", "#E7B800", "purple", "Green"),
        legend.title = "",
        risk.table.y.text.col = TRUE,
        risk.table.height = 0.3
      ) 
      pdf(paste(outfile, Xin, "KM.pdf", sep = "_"))
      print(newplot)
      dev.off()
      newplot
    }
  }
  
  fit1 = survival::coxph(as.formula(paste(survY, survX, sep=" ~ ")), data = datain)
  
  coxphObj = summary(fit1)
  
  ### focus on $coef, $conf.int
  coxcoef = coxphObj$coefficients
  coxci = coxphObj$conf.int
  
  coxout = coxcoef
  
  mm = dim(coxout)[2]
  
  if(dim(coxout)[1] == 1){
    tmp1 = as.numeric(coxout)
    tmp2 = as.numeric(coxci)
    coxout = c(tmp1[1:2], tmp2[3:4], tmp1[(mm-2):mm])
    coxout = data.frame(matrix(coxout, nrow = 1, ncol = 7))
    
  }else{
    coxout = cbind(coxout[,c(1:2)], coxci[,3:4],coxout[,(mm-2):mm])
  }
  colnames(coxout) = c("beta", 'HR', "HR_95%CI_lower","HR_95%CI_upper", "beta_se", "beta_z", "Pvalue_beta")
  
  return(list(fit1,coxout, newplot))
}