You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This commit was created on GitHub.com and signed with GitHub’s verified signature.
Bug fixes
check_sum_of_transition_array() (R/Functions.R) compared the sum of each
cycle's transition probabilities in a transition array with the number of
states using exact equality, so rounding error could make it reject a valid
array. It now uses a numerical tolerance (all.equal()), as in darthtools.
This tutorial only checks transition matrices, so its results are not
affected; the same bug stopped the PSA of the state-residence model in cohort-modeling-tutorial-timedep
(fixed there in v0.3.0).
Documentation
Documented an erratum in the published article's parameter table (Table 2),
which lists the cost of treatment B (c_trtB) as $12,000 instead of the
$13,000 stated in the text and used in the code, the PSA and the published
results. The code was already correct. See manuscript/README.md and the
README (closes #1). Reported by @conallaque.
Tests
Added tests/testthat/test-psa-params.R, which checks that every PSA
distribution is centred on its base-case value (within 2%), so the base
case and the PSA cannot drift apart.
Added tests/testthat/test-transition-checks.R for the transition matrix
and array checks.