Bayes, prevalence, & PPV
While I was taking a course (online) about Bayesian statistics, one of the intro sections had an example of calculating the positive predictive value of a screening test using Bayes theorem. This reminded me of some really old posts that I came across in college particularly regarding screening tests (e.g. mammography).
I’m a big fan of interactive apps to display principles, so I took a shot at developing my own. This Shiny app shows exactly how sensitive the positive predictive value is to the underlying prevalence. Granted this is something that I knew conceptually, but actually being able to interact with the data really reinforces the degree of skepticism we should have towards positive screening tests in populations with a low prevalence.
Below is an iframe of the app, but I’d recommend visiting the full app if you want to play around with it.