{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiejz43l54blse5bvl665zaql7exnopuvqcjolfranjwammbti46ra",
    "uri": "at://did:plc:wwyqal4cnqhuwyacdj7rqq3n/app.bsky.feed.post/3mkkrez5s55s2"
  },
  "path": "/t/rms-semiparametric-ordinal-longitudinal-model/4819?page=6#post_115",
  "publishedAt": "2026-04-28T13:01:27.000Z",
  "site": "https://discourse.datamethods.org",
  "textContent": "That’s a somewhat unusual but extremely interesting example. I hope the dataset can be used by others to explore other things. The first thing to explore is to try to find utilities for all the states and to see which treatment has the highest expected utility by 28d.\n\nI wouldn’t say this is a power reduction per se, but rather an averaging out of the effect over many outcome levels.\n\nA Bayesian design would cover the bases, e.g., success criteria could be P(OR for mortality < 1) > 0.95 or P(overall OR < 0.85) > 0.95.",
  "title": "RMS Semiparametric Ordinal Longitudinal Model"
}