library(Hmisc)
#mtd <- read.csv("../data/SI-NET-MTD-updated for stat.csv")
mtd <- read.csv("../data/SINET-MTD-stats-04-02-13.csv") # Three patients (mrn = c(11, 12, 31)) have been switched from DOD to D other.
#mtd <- read.csv("../data/newSI-NET-MTD.csv")

mtd <- upData(mtd, lowernames = TRUE,
   drop = c("x", "x.1", "x.2"),
   rename = list(m.f = "gender",
      lesion.size = "tumorSize",
      lesion.number = "numberTumors",
      ln = "lymphNodeMet",
      dos = "surgeryDate",
      liver = "liverMet",
      t.stage = "tStage",
      fu = "followupDate",
      x.liver.lesions.at.surgery = "numLiverLesSurg",
      liver.progression = "liverProgression",
      dod = "deathDate",
      date.of.progression = "progressionDate",                  # Is this only progressions after surgery??
      date.of.1st.liver.lesion = "firstLiverLesionDate",
      a.d = "status"),
   tumorSize = as.numeric(as.character(tumorSize)),
   multipleLesions = factor(numberTumors),
   lymphNodeMet = factor(lymphNodeMet),
   liverMetAtSurgery = factor(numLiverLesSurg),
   mtd = factor(mtd),
   liverMet = factor(liverMet),
   lvi = factor(lvi),
   tStageCat = factor(tStage),
   logTumorSize = log(tumorSize),
   surgeryDate = as.Date(as.character(surgeryDate), format = "%m/%d/%Y"),
   followupDate = as.Date(as.character(followupDate), format = "%m/%d/%Y"),   # per Chanjuan, follow up date is date of death for those who died.
   progressionDate = as.Date(as.character(progressionDate), format = "%m/%d/%Y"),
   deathDate = as.Date(as.character(deathDate), format = "%m/%d/%Y"),
   firstLiverLesionDate = as.Date(as.character(firstLiverLesionDate), format = "%m/%d/%Y"),
   osTime = as.numeric(followupDate - surgeryDate)*12/365.25,
   firstLiverLesionTimeSurg = as.numeric(firstLiverLesionDate - surgeryDate)*12/365.25,
   osStatus = factor(status),
   osStatusCensorDOC = factor(status),
   crStatus = factor(status),
   crPfsStatus = factor(ifelse(status %in% c("ANED", "AWD") & liverProgression %in% "0", "Progression-free, alive", 
      ifelse(status == "DOD" | liverProgression %in% "1", "Progressed or died of disease", 
         ifelse(status == "D other", "Died of other causes", NA)))),
   levels = list(gender = list("Male" = "M", "Female" = "F"),
      osStatus = list("Alive" = c("ANED", "AWD"), "Dead" = c("DOD", "D other")),
      osStatusCensorDOC = list("Alive or censored" = c("ANED", "AWD", "D other"), "Dead" = "DOD"),
      crStatus = list("Alive" = c("ANED", "AWD"), "Dead: disease" = "DOD", "Dead: other cause" = "D other"),
      crPfsStatus = list("Progression-free, alive" = "Progression-free, alive", "Progressed or died of disease" = "Progressed or died of disease", "Died of other causes" = "Died of other causes"), #need to check how to treat AWD and liverProgression
      tStageCat = list("T-stages 1, 2" = c("1", "2"), "T-stages 3, 4" = c("3", "4")),
      liverMet = list("No" = "0", "Yes" = "1"),
      liverProgression = list("No" = "0", "Yes" = "1"),
      mtd = list("No MTD" = "0", "MTD" = "1"),
      lvi = list("No LVI" = "0", "LVI" = "1"),
      multipleLesions = list("One" = "1", "Multiple" = c("2", "3", "4", "5", "6", "7", "multiple")),
      liverMetAtSurgery = list("No" = "0", "Yes" = c("1", "2", "3", "4", "5", "multiple", "numerous")),
      lymphNodeMet = list("No lymph node met" = "0", "Lymph node met" = "1")),
   labels = list(gender = "Gender",
      age = "Age",
      tumorSize = "Tumor size",
      logTumorSize = "Log tumor size",
      numberTumors = "Number of tumors",
      multipleLesions = "Number of tumors",
      lymphNodeMet = "Lymph node metastasis",
      mtd = "Mesenteric tumor deposits",
      liverMet = "Liver metastasis",
      liverMetAtSurgery = "Liver metastasis at time of surgery",
      liverProgression = "Liver progression after surgery",
      lvi = "Lymphovascular invasion",
      tStage = "T stage",
      firstLiverLesionTimeSurg = "Time from surgery to first liver lesion (months)",
      status = "Status",
      tStageCat = "T stage",
      osTime = "Overall survival follow up time (months)",
      crStatus = "Survival status",
      crPfsStatus = "Progression-free survival status",
      osStatus = "Overall survival status"))



mtd <- upData(mtd,
   mtdLnm = interaction(mtd, lymphNodeMet),
   crStatusNum = as.numeric(crStatus) - 1,
   crPfsStatusNum = as.numeric(crPfsStatus) - 1,
   pfsStatus = factor(osStatus %in% "Dead" | liverProgression %in% "Yes"),
   pfsStatusCensorDOC = factor(osStatusCensorDOC %in% "Dead" | liverProgression %in% "Yes"),
   pfsTime = as.numeric(as.Date(ifelse(liverProgression == "Yes", progressionDate, followupDate), origin = "1970-01-01") - surgeryDate)*12/365.25,
   levels = list(pfsStatus = list("Progression-free, alive" = "FALSE", "Progressed or died" = "TRUE"),
      pfsStatusCensorDOC = list("Progression-free and alive or DOC," = "FALSE", "Progressed or died of disease" = "TRUE")),
   labels = list(pfsStatus = "Progression-free survival status",
      pfsStatusCensorDOC = "Progression-free survival status",
      pfsTime = "Progression-free survival follow up time (months)"))