Hacktoberfest 2026: le issue che i maintainer hanno segnato per ottobre, aperte e adatte ai principianti. Sfoglia le issue Hacktoberfest

Add Survival Mixture Density Networks for survival prediction

Aperta
  • Qualcosa non va?
#1,297 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 3 giorni

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
4/5
Tempo stimato
3-5 giorni
Idoneità per principianti
30/100
Tipo di issue
Funzionalità
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
python
Ambito
machine-learning

Direzione di ricerca

Start by reading the existing BaseModel and the trainer's custom-loss support, since the issue plans to build on both. The first step is a SurvivalMDN model with a right-censored likelihood loss, checked against the authors' implementation on controlled inputs. Done means the model's loss and predictions at chosen times are tested, then the SUPPORT2 task and an end-to-end example are added. Maintainer feedback on scope is requested first, so confirm the plan before writing code.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

We’d like to contribute an implementation of Survival Mixture Density Networks (Han et al., 2022) to PyHealth as part of a UIUC course project.

We propose adding a reusable Survival MDN model for continuous-time survival prediction with right-censored data. The model would accept patient features, observed follow-up times, and event indicators from compatible datasets. A dataset-specific SUPPORT2 task and an end-to-end example would demonstrate its use.

We plan to add:

  • A SurvivalMDN model extending BaseModel, with a censored likelihood loss and survival predictions at user-selected times.
  • Patient-specific, learned shared, and uniform mixture-weight options.
  • A SUPPORT2 time-to-event task.
  • Tests, documentation, and a training/evaluation example.

We’ll validate loss values and predictions against the authors’ implementation on controlled inputs and use their released SUPPORT benchmark for our reproduction. We’ll write an original implementation based on the paper, with attribution.

For training, we plan to use the existing trainer’s custom-loss support and add a separate survival evaluation utility. We’ll develop this in small, tested steps, starting with the model and then adding the SUPPORT2 task and evaluation example.

We’d welcome feedback on the proposed scope or any overlapping work.

Lingua principale
Python
Stelle
1.7k
Fork
803
Merge medio
5g 20h
PR unite (30g)
56

Preparare l'ambiente

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di sunlabuiuc/PyHealth

Tutte le issue di sunlabuiuc/PyHealth

Issue simili

Altre issue su Python

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.