Optimise the capacity profile for projections

Usage

optimise_capacity(
  t_1_capacity,
  referrals_projections,
  incomplete_pathways,
  renege_capacity_params,
  target,
  target_bin,
  capacity_profile = "linear_change",
  surplus_treatment_redistribution_method = "evenly",
  tolerance,
  max_iterations = 50
)

Arguments

t_1_capacity: numeric; a single value for the capacity for the first time period of the projected time period
referrals_projections: numeric; a vector for the number of referrals for each period in the projected time period
incomplete_pathways: tibble; two column data frame or tibble, with fields called months_waited_id (taking values 0 to the maximum months waited group of interest), and incompletes (the count of the number of incomplete pathways) representing the count of incomplete pathways at timestep 0 (to initialise the model with)
renege_capacity_params: tibble; three column data frame or tibble, with fields called months_waited_id (taking values 0 to the maximum months waited group of interest), and fields called capacity_param and renege_param, which are outputs from the function calibrate_capacity_renege_params()
target: string length 1; can be either a percentage point change, eg, "~-5%" or a percent value, eg, "5%". This refers to percentage of patients on the waiting list in the target_bin or higher waiting times. Note, this is the opposite of the NHS RTT targets, which are a proportion of patients on the waiting list that are below the target_bin
target_bin: numeric length 1; the bin that the target refers to. It must be less than or equal to the max_months_waited value
capacity_profile: string, one of "linear_change" or "flat"; determines how the capacity counts vary into the future. Linear change means that the first point is held stationary at the value of t_1_capacity and the end point is varied, with a linear interpolation between the two points. Flat means that capacity remains constant into the future
surplus_treatment_redistribution_method: string; one of "none", "evenly" or "prioritise_long_waiters"; should cases where the counts of reneges and treatments exceed the counts of people waiting be redistributed, and if so, which method should be used
tolerance: numeric length 1; the tolerance used to compare the absolute error with in the max_months_waited bin to determine convergence. The absolute error is calculated on the proportion in the max_months_waited bin relative to the total waiting (even if a non-percentage target is used)
max_iterations: numeric; the maximum number of iterations to test for convergence before providing a warning and an invalid number

Value

A capacity multiplier representing that is applied to the t_1_capacity input to achieve the desired target by the end of the projection period. Where the capacity_profile is 'linear_change', this represent a linear growth in capacity from t₁ to t₁₃, where t₁ is equal to t_1_capacity and t₁₃ is t_1_capacity multiplied by the output of the function.

If capacity_profile is 'flat', then the projected capacity is simply t_1_capacity multiplied by the output of the function for the whole of the projected period.

The name of the returned object provides an indication of whether the optimiser converged.