{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreih7wnlo6txxf6fzgztmosywr5x6run2tr4ddereemo45iclmhwxli",
    "uri": "at://did:plc:wwyqal4cnqhuwyacdj7rqq3n/app.bsky.feed.post/3mhynsmpm3n72"
  },
  "path": "/t/best-model-reduction-approach-for-fine-gray-prediction-model/28678#post_4",
  "publishedAt": "2026-03-26T12:22:10.000Z",
  "site": "https://discourse.datamethods.org",
  "tags": [
    "https://cran.r-project.org/web/packages/survival/vignettes/compete.pdf",
    "Discrete time multistate models",
    "https://discourse.datamethods.org/t/clinical-trial-outcomes-interrupted-by-other-outcomes"
  ],
  "textContent": "Yes if there are many candidate predictors then unsupervised learning is a great first step.\n\nRegarding the big picture, I don’t find prediction of AD that precedes death to be that meaningful. That’s what competing risk analysis does. I find it much more natural to use multi-state models. See [this](https://cran.r-project.org/web/packages/survival/vignettes/compete.pdf) amazing document by Therneau, Crowson, Atkinson.\n\nDiscrete time multistate models are even simpler.\n\n[This survey](https://discourse.datamethods.org/t/clinical-trial-outcomes-interrupted-by-other-outcomes) revealed that most researchers find it impossible to really separate death from nonfatal outcomes anyway.",
  "title": "Best model reduction approach for Fine-Gray prediction model"
}