This model agnostic function runs a single markov model for the specified duration using a Armadillo implementation.
See example_two_state_markov for an example of the required input. Alternatively use sample_markov(type = "base")
and the output from sample_markov.
ArmaSimulateMarkov( sim, cohort, transition, duration, state_cost, discounting, qalys, intervention_cost )
| sim | Matrix with the same number of rows as the duration of the model and the same number of columns as the number of states in the model. Used to store model simulations. |
|---|---|
| cohort | A vector containing the initial state. |
| transition | A transition matrix, see |
| duration | Numeric, how many long to run the model for. |
| state_cost | A list of state costs for each intervention,
see |
| discounting | Numeric vector, the discount that should be applied to the costs and QALYs for each time period.
This must be the same length as |
| qalys | A list of QALYs for each intervention, see |
| intervention_cost | A vector of intervention costs, see |
A list containing total costs and total QALYs as matrices across states and the duration of the model
## Sample model markov_sample <- sample_markov(example_two_state_markov()) ## Simulate using R sim_r <- simulate_markov_base( ## Specify the storage simulation matrix to maintain consistency ##here (but not needed for the base implementation). sim = matrix(NA, nrow = 10, ncol = nrow(markov_sample$transition[[1]])), transition = markov_sample$transition[[1]], cohort = markov_sample$cohort[[1]], state_cost = markov_sample$state_cost[[1]], intervention_cost = markov_sample$intervention_cost[[1]], qalys = markov_sample$qalys[[1]], duration = 10, discounting = SpeedyMarkov::calc_discounting(1.035, 10), markov_loop_fn = SpeedyMarkov::markov_loop ) # RcppArmadillo implementation sim_rcppArma <- ArmaSimulateMarkov( sim = matrix(NA, nrow = 10, ncol = nrow(markov_sample$transition[[1]])), transition = markov_sample$transition[[1]], cohort = markov_sample$cohort[[1]], state_cost = markov_sample$state_cost[[1]], intervention_cost = markov_sample$intervention_cost[[1]], qalys = markov_sample$qalys[[1]], duration = 10, discounting = SpeedyMarkov::calc_discounting(1.035, 10) ) # Check results are within tolerances all.equal(sim_r, sim_rcppArma)#> [1] TRUE# Benchmark library(microbenchmark) microbenchmark(simulate_markov_base( sim = matrix(NA, nrow = 100, ncol = nrow(markov_sample$transition[[1]])), transition = markov_sample$transition[[1]], cohort = markov_sample$cohort[[1]], state_cost = markov_sample$state_cost[[1]], intervention_cost = markov_sample$intervention_cost[[1]], qalys = markov_sample$qalys[[1]], duration = 100, discounting = SpeedyMarkov::calc_discounting(1.035, 100), markov_loop_fn = SpeedyMarkov::markov_loop), ArmaSimulateMarkov( sim = matrix(NA, nrow = 100, ncol = nrow(markov_sample$transition[[1]])), transition = markov_sample$transition[[1]], cohort = markov_sample$cohort[[1]], state_cost = markov_sample$state_cost[[1]], intervention_cost = markov_sample$intervention_cost[[1]], qalys = markov_sample$qalys[[1]], duration = 100, discounting = SpeedyMarkov::calc_discounting(1.035, 100)), times = 1000)#> Unit: microseconds #> expr #> simulate_markov_base(sim = matrix(NA, nrow = 100, ncol = nrow(markov_sample$transition[[1]])), transition = markov_sample$transition[[1]], cohort = markov_sample$cohort[[1]], state_cost = markov_sample$state_cost[[1]], intervention_cost = markov_sample$intervention_cost[[1]], qalys = markov_sample$qalys[[1]], duration = 100, discounting = SpeedyMarkov::calc_discounting(1.035, 100), markov_loop_fn = SpeedyMarkov::markov_loop) #> ArmaSimulateMarkov(sim = matrix(NA, nrow = 100, ncol = nrow(markov_sample$transition[[1]])), transition = markov_sample$transition[[1]], cohort = markov_sample$cohort[[1]], state_cost = markov_sample$state_cost[[1]], intervention_cost = markov_sample$intervention_cost[[1]], qalys = markov_sample$qalys[[1]], duration = 100, discounting = SpeedyMarkov::calc_discounting(1.035, 100)) #> min lq mean median uq max neval #> 112.193 114.681 128.06289 116.1990 123.5925 3544.577 1000 #> 29.562 30.957 33.39825 31.6415 33.5150 184.803 1000