ictstm$sim_disease() function error: Error: Expecting a single value: [extent=0].

[mcincam100]

Hi,
I am doing some practice runs of a flexsurv + hesim simulation, using artificial data. All seems to go well until I hit the ictstm$sim_disease() function, and I get an error:

ictstm$sim_disease()
Error: Expecting a single value: [extent=0].
My full code is below. I have been unable to successfully diagnose the issue. Thanks for any tips!
Cameron

#STEP 1: CREATING HESIM_DAT FILE FROM MULTIPLE CSVs CONTAINING PIECES OF DISEASE PROGRESSION #FRAMEWORK, TO #INITIALIZE THE SIMULATION

Load necessary libraries

library(hesim)

Write strategies.csv

strategies <- data.frame(strategy_id = c(0, 1))
write.csv(strategies, "J:/Data Development Unit/Administrative/Analytical Diversification/R Testing/hesimstrat.csv", row.names = FALSE)

Write patients.csv

set.seed(123) # for reproducibility
patients <- data.frame(patient_id = seq(1, 1000),
x1 = rbinom(1000, 1, .5),
x2 = rbinom(1000, 1, .5),
x3 = rbinom(1000, 1, .5))
write.csv(patients, "J:/Data Development Unit/Administrative/Analytical Diversification/R Testing/hesimpat.csv", row.names = FALSE)

Write states.csv

states <- data.frame(state_id = c(1, 2, 3, 4))
write.csv(states, "J:/Data Development Unit/Administrative/Analytical Diversification/R Testing/hesimstat.csv", row.names = FALSE)

Read the .csv files into R

strategies <- read.csv("J:/Data Development Unit/Administrative/Analytical Diversification/R Testing/hesimstrat.csv")
patients <- read.csv("J:/Data Development Unit/Administrative/Analytical Diversification/R Testing/hesimpat.csv")
states <- read.csv("J:/Data Development Unit/Administrative/Analytical Diversification/R Testing/hesimstat.csv")

Define transitions

tmat <- rbind(c(NA, 1, NA, NA),
c(2, NA, 3, NA),
c(4, NA, NA, 5),
c(6, 7, NA, NA))
colnames(tmat) <- rownames(tmat) <- c("nonCJS", "Police", "Courts", "Corrections")
transitions <- create_trans_dt(tmat)
transitions[, trans := factor(transition_id)]

Create the hesim_dat object

hesim_dat <- hesim_data(strategies = strategies,
patients = patients,
states = states,
transitions = transitions)

Print the hesim_dat object

print(hesim_dat)

#Expand the hesim_dat object into long format
transmod_data <- expand(hesim_dat,
by = c("strategies", "patients", "transitions"))

options(max.print = 10000) # Increase max.print to a large number
head(transmod_data, 14) # Now this should print 14 rows (can be changed to any desired number)

#STEP 2: CREATING FLEXSURV MSTM FIT OBJECT TO ENABLE THE DISEASE PROGRESSION MODEL IN HESIM #(WEIBULL AFT CLOCK-RESET MODEL)

Load necessary libraries

library(flexsurv)
library(survival)

Set the seed for reproducibility

set.seed(123)

Generate data for flexsurv analysis

n <- 1000 # Define the number of patients
patient_id <- rep(1:n, each = 7) # Each patient has 7 transitions
time <- rweibull(n * 7, shape = 2, scale = 1) # Modify parameters as needed
status <- rbinom(n * 7, size = 1, prob = 0.5)
trans <- rep(1:7, times = n) # 7 transitions
strategy_id <- rep(rbinom(n, size = 1, prob = 0.5), each = 7)
x1 <- rep(rbinom(n, size = 1, prob = 0.5), each = 7)
x2 <- rep(rbinom(n, size = 1, prob = 0.5), each = 7)
x3 <- rep(rbinom(n, size = 1, prob = 0.5), each = 7)

Create data frame

data <- data.frame(patient_id, time, status, trans, strategy_id, x1, x2, x3)

Write to CSV

write.csv(data, file = "J:/Data Development Unit/Administrative/Analytical Diversification/R Testing/flexsurvmod.csv", row.names = FALSE)

Define your transition data

surv_dat <- read.csv("J:/Data Development Unit/Administrative/Analytical Diversification/R Testing/flexsurvmod.csv")

Fit the survival models

fits <- vector(length = max(tmat, na.rm = TRUE), mode = "list")
for (i in 1:length(fits)){
fits[[i]] <- flexsurvreg(Surv(time, status) ~ factor(strategy_id),
data = surv_dat,
subset = (trans == i),
dist = "weibull")
}
fits <- flexsurvreg_list(fits)

#STEP 3: CREATING/RUNNING THE DISEASE PROGRESSION MODEL IN HESIM

transmod_data <- expand(hesim_dat)
transmod <- create_IndivCtstmTrans(fits, input_data = transmod_data,
trans_mat = tmat,
n = 200)
class(transmod)

ictstm <- IndivCtstm$new(trans_model = transmod,
utility_model = NULL,
cost_models = NULL)

ictstm$sim_disease()
head(ictstm$disprog_)

[dincerti]

The problem is that your transition model doesn't have a death state specified. Try this:

transmod <- create_IndivCtstmTrans(
  fits,
  input_data = transmod_data,
  trans_mat = tmat,
  n = 200,
  death_state=4
)

The reason no death state is specified is because your tmat doesn't have any absorbing states so a death state cannot be automatically determined. That said, this is a bug: a much better error message should be displayed alerting you of the problem.

[mcincam100]

Many thanks Devin, that worked.
Cam