search

audrey-b / BUGSnet

BUGSnet: a new comprehensive package for network meta-analysis
Other
23 stars 9 forks source link

UBC improvements suggestions #54

Open audrey-b opened 5 years ago

audrey-b commented 5 years ago

data plot increase the font (too small when lots of data) DONE plot single arm studies (currently net.plot gives error) bias adjustment models leverage : which arms they come from PROBABLY EASY sort forest plot DONE get indirect estimate color edges by comparison (high vs low vs unclear risk of bias) weight by number of patients help file: weight is by number of studies in network plot DONE help file: se is sd/sqrt(n) not n DONE meta-regression with categorical (low, high, unclear risk of bias)

audrey-b commented 5 years ago

Jon:

  1. Change documentation on BugsNet re: data.plot for nodes/edges (ie: change size/colour related to other categorical/continuous variables)
  2. Identification of study and arm when interpreting posterior mean deviance (consistency/inconsistency) plots
  3. Identification of indirect effect estimates for pairwise comparisons when available (for comparison vs direct estimates)
  4. Meta-regression with continuous or categorical variables, and for more than a single variable at a time (and potentially for interactions?)

Ben: -It would be great if BUGSnet could produce similar results to CiNeMa but using the Bayesian results. -Give estimates for both direct, indirect, and Bayesian network estimates.