FRA-RIG LUNG Cancer (TCGA-LUAD, sklearn demo): survival risk models

TL;DR • Two survival models for TCGA-LUAD (PanCancer Atlas): (A) outcome-aware (v4) — CV ROC-AUC ≈ 0.93 (leak-prone, for study only) (B) honest baseline (v4.1) — CV ROC-AUC ≈ 0.76 (no outcome leakage; deploy-style). • 1-minute run: pip install -r requirements.txt → python predict_demo.py --model v4_1 --csv sample.csv • Open code: notebook + ZIP. Data: cBioPortal (TCGA-LUAD PanCancer Atlas). Models/reports: CC BY 4.0. Code: MIT. Research only, not for clinical use.