Skip to content

Ra/labs only variant - #41

Draft
will-pang wants to merge 21 commits into
mainfrom
ra/labs-only-variant
Draft

will-pang wants to merge 21 commits into
mainfrom
ra/labs-only-variant

Conversation

@will-pang

Copy link
Copy Markdown

Description

Overview of Changes

Testing Notes

  • {You should have tested the code in some way (doesn't have to be a unit test)}
  • {Describe any additional tests}

rianfyi and others added 20 commits May 14, 2026 05:44
Pure lab-value mortality prediction (no notes, no ICD codes). Mirrors
the labs/labs_mask schema from NotesLabsMIMIC4 so the same backbones
work unchanged. Full-scale results: transformer 0.496/0.888 PR/ROC-AUC
at ep9, MLP 0.399/0.880 at ep38 — best non-leaky result across 18 runs.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…into ra/labs-only-variant

# Conflicts:
#	examples/mortality_prediction/unified_embedding_e2e_mimic4.py
#	pyhealth/models/embedding/unified.py
#	pyhealth/tasks/multimodal_mimic4.py
#	scripts/slurm/run_table2.sh
@rianfyi

rianfyi commented Aug 13, 2026

Copy link
Copy Markdown

Closing as a duplicate of #33 on the same branch (ra/labs-only-variant). The labs-only baseline that should land is in #46 (LabsOnlyMIMIC4 plus the observation-window fix).

@rianfyi rianfyi closed this Aug 13, 2026
@rianfyi

rianfyi commented Aug 13, 2026

Copy link
Copy Markdown

Reopened. Closing this one is Will's call; I should not have closed someone else's PR.

@rianfyi rianfyi reopened this Aug 13, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants