06 progressDisease.r

##################################################
## Allen Roberts
## July 7, 2015
## Edited by Jessica Culhane on 9/20/18
## Description: Disease progression function (CD4 count and viral load)
##################################################

## This function takes the population data.table as input and progresses HIV+ persons (not on ART) through CD4 
## and viral load categories. Note that it modifies dt by reference, which points to the data.table that was 
## originally passed to it.  This means that it's modifying that data.table in the parent environment.

progressDisease <- function(dt, time_step) {
  
  ## Note that we want the probability of progressing in a given time step.  This can be obtained from the mean 
  ## duration using the exponential decay function (assuming a constant rate).  
  ## http://hyperphysics.phy-astr.gsu.edu/hbase/nuclear/meanlif.html
  
  setkey(dis_prog, hiv, male, vl, cd4, vs)
  setkey(dt, hiv, male, vl, cd4, vs)
  dt[dis_prog, c("cd4_duration", "vl_duration") := list(cd4_duration, vl_duration)]
  
  ## CD4 Efflux
  dt[!is.na(cd4_duration), diff := diff - count * (1 - exp(-time_step/cd4_duration))]
  
  ## VL Efflux
  dt[!is.na(vl_duration), diff := diff - count * (1 - exp(-time_step/vl_duration))]
  
  ## Influx - need to get counts from previous CD4 and VL categories, calculate how many people leave the 
  ## category, increment the category by 1, and merge back to the original data frame and add that number 
  ## of people.  
  ## CD4 Influx
  prev_cd4 <- copy(dt[vl != 6 & cd4 >= 1 & cd4 < 5 & hiv == 1 & vs == 0])
  prev_cd4[, cd4_influx := count * (1 - exp(-time_step/cd4_duration))]
  prev_cd4[, cd4 := cd4 + 1]
  prev_cd4[, c("count", "cd4_duration", "vl_duration", "diff") := NULL]
  setkeyv(prev_cd4, all_keys)
  setkeyv(dt, all_keys)
  dt[prev_cd4, cd4_influx := cd4_influx]
  dt[!is.na(cd4_influx), diff := diff + cd4_influx]
  dt[, cd4_influx := NULL]
  
  ## VL Influx
  prev_vl <- copy(dt[cd4 != 6 & vl >= 1 & vl < 5 & hiv == 1 & vs == 0])
  prev_vl[, vl_influx := count * (1 - exp(-time_step/vl_duration))]
  prev_vl[, vl := vl + 1]
  prev_vl[, c("count", "cd4_duration", "vl_duration", "diff") := NULL]
  setkeyv(prev_vl, all_keys)
  setkeyv(dt, all_keys)
  dt[prev_vl, vl_influx := vl_influx ]
  dt[!is.na(vl_influx), diff := diff + vl_influx]
  dt[, c("vl_influx", "cd4_duration", "vl_duration") := NULL]
  
}