- partial argument matching in https://github.com/chjackson/voi/blob/master/R/evppi_gam.R#L11. nicer to make inla and gp methods return classed model object and use fitted() method.
-
More built-in common study designs for EVSI, e.g. binary trial that just informs the odds ratio.
-
Different regression methods and options for effects and costs (e.g. int.ord)
Features that I don't plan to implement myself, but would accept cleanly-written contributions for (see coding_style.md)
-
Parallel processing for 2 level Monte Carlo, and partial results for convergence checking
-
Gaussian Approximation EVSI method
-
MLMC and QMC methods
-
2-level Monte Carlo EVSI
-
Check can reproduce results from ConVoI 1 four case studies
-
Standard errors for EVSI methods and INLA EVPPI method