Binomial Distributions

In creating a binomial distribution for a PSA, do the "Trials" equal the number of stages in your Markov model? As well, in what circumstances would the "p(probability)" differ from the "override mean"? I want to use the binomial distribution for selected probabilities as  well as costs. Thank you



1 comment
  • Official comment

    In most cases "trials" in the context of TreeAge Pro correspond to number of patients simulated during microsimulation (individual patient simulation).  The "stage" corresponds to time interval that your model is supposed to simulate (e.g. stage = month of calendar time).  When such monthly model runs for 12 stages (a.k.a cycles) it simulates passage of 1 year.

    However, in the case of Binomial Distribution parameters the term "Trial" has a more traditional probabilistic meaning or how many times a coin or dice was thrown.  So in your case if you want to draw a random number corresponding to the number of "successes" over the number of _stages, you could define your distribution as you suggest for the PSA analysis. 

    You do not have to override the mean, but you have that option if you want your base case to be referenced from a specific value rather than the calculated distribution mean.  For example if you have highly skewed distribution, you might want to use a different value than the mean (for example a median), however, the usage of mean override is rather rare.

    Feel free to reach out to to discuss any additional questions regarding your use of the Binomial distribution.

Please sign in to leave a comment.

Didn't find what you were looking for?

New post