使用deSolve包-ddply函数模拟数据

Question

I have trouble applying the differential equation solver for each subject ID in the deSolve package to calculate drug amount in a 2-compartment intravenous infusion model. I was able to set the code (below) so it solves for one subject. I need help in how to apply it for every subject in the data frame using the ddply package.

Below is a data frame called simeventdfi that has the dose events needed for calculations.

library(deSolve)

Step1: provide the event df for each subject

 simeventdfi <- 
 ID  var time value method
 1    3   0.0   6   rep
 1    3  16.7   0   rep
 1    4  0.0    2.4 rep
 1    4  16.7   2.4 rep
 1    5  0.0    10  rep
 1    5  16.7   10  rep
 1    6  0.0    1   rep
 1    6  16.7   1   rep
 1    7  0.0    25  rep
 1    7  16.7   25  rep
 2    3  0.0    6   rep
 2    3  16.7   0   rep
 2    4  0.0    2.4 rep
 2    4  16.7   2.4 rep
 2    5  0.0    10  rep
 2    5  16.7   10  rep
 2    6  0.0    1   rep
 2    6  16.7   1   rep
 2    7  0.0    25  rep
 2    7  16.7   25  rep

Step 2: Specify the times for the simulated values

   simtimes <- sort(unique(c(simeventdfi$time, seq(0,10,1))))

Step 3: Function containing differential equations for amounts in compartments (A)

  #THETAin is empty for this model   
   simthetai <- NULL

 DES <- function(T, A, THETAin)
  {

    RATE  <- A[3]    #Dose rate (time dependent)
    CL <- A[4]         #Time dependent                           
    V1 <- A[5]         #Time dependent
    Q  <- A[6]         #Time dependent
    V2 <- A[7]         #Time dependent

    dA1 <- RATE -Q/V1*A[1] +Q/V2*A[2] -CL/V1*A[1]  #Central compartment 
    dA2 <- Q/V1*A[1] - Q/V2*A[2]     #Peripheral compartment 

    RATE <- 0  #Set rate to zero so doesn't change unless event
    dCL <- 0 
    dV1 <- 0
    dQ <- 0
    dV2 <- 0   
   list(c(dA1,dA2,RATE,dCL,dV1,dQ,dV2))   #List of derivatives
     }

Step 4: Since I was able to set the functioncode to solve for one subject ID only, then I have to subset one subject ID from the simeventdfi given above. However, I want help in how to apply the differential equation solver for every subject ID using ddply . For now, I will subset the simeventdfi above for one subject to demonstrate how the function works.

simeventdfi <- subset(simeventdfi,ID==1)

Set initial values - compartments and time-dependent parameters

    A_0i <- c("A1"=0,"A2"=0,
      "Rate"=simeventdfi$value[1],
      "CL"=simeventdfi$value[simeventdfi$var==4 & simeventdfi$time==0],
      "V1"=simeventdfi$value[simeventdfi$var==5 & simeventdfi$time==0],
      "Q"= simeventdfi$value[simeventdfi$var==6 & simeventdfi$time==0],
      "V2"=simeventdfi$value[simeventdfi$var==7 & simeventdfi$time==0]) 
   print(A_0i)

Step 5: Run differential equation solver and get results in a dataframe

    simdatadfi <- as.data.frame(ode(A_0i, simtimes, DES, simthetai, events=list(data=simeventdfi), method="lsoda"))

It runs fine with one subject only. I need help in how to apply the differential equation solver in Step 5 for each subject in simeventdfi .

Answer 1

You can use dplyr . Just use group_by(ID) to apply the differential equation for each subject. Apparently ode does not like the dplyr data format, so you need to use data.frame(.) instead of just . in the do call.

require(dplyr)
simeventdfi %>% 
  group_by(ID) %>% 
  do(
    ode(A_0i, 
        simtimes, 
        DES, 
        simthetai, 
        events=list(data=data.frame(.)), 
        method="lsoda") %>%
      data.frame
  ) %>% 
  data.frame

Answer 2

To apply the differential equation solver to every subject:

First: write step 4 as a function:

  simulate.conc <- function(simeventdfi) { 

      #Initial values - compartments and time-dependent parameters
      A_0i <- c("A1"=0,"A2"=0,
        "Rate"=simeventdfi$value[1],
        "CL"=simeventdfi$value[simeventdfi$var==4 & simeventdfi$time==0],
        "V1"=simeventdfi$value[simeventdfi$var==5 & simeventdfi$time==0],
        "Q"= simeventdfi$value[simeventdfi$var==6 & simeventdfi$time==0],
        "V2"=simeventdfi$value[simeventdfi$var==7 & simeventdfi$time==0]) 

   #Run differential equation solver 
     simdata <- ode(A_0i, simtimes, DES, simthetai, events=list(data=simeventdfi), method="lsoda")

 #Process the simulated output 
  simdata <- as.data.frame(simdata)

  #Concentration in the central compartment
  simdata$conc <- simdata$A1/simdata$V1    #Amount to concentration

  #remove any duplicated datapoints
  simdata <- simdata[!duplicated(simdata), ]
 }

Second: apply the function on every ID in the simeventdfi using ddply

 library(plyr)
 simdata_DES <- ddply(simeventdfi, .(ID), simulate.conc)

Easy !