GRASP: GRoup-shApley feature Selection for Patients
Yuheng Luo, Shuyan Li, Zhong Cao · 2026
Read GRASP: GRoup-shApley feature Selection for Patients by Yuheng Luo, Shuyan Li, Zhong Cao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-11084.txt.
Description
Medical forecast needs trait sets that are small, stable, and readable. Yuheng Luo and two co-authors offer GRASP, which pairs Shapley attribution with group sparsity to pick such sets. Importance scores are distilled from a pretrained tree model, then a group-regularized logistic model enforces structured sparsity.
Compared with LASSO, SHAP, and deep methods, GRASP matches or beats predictive scores while using fewer, less redundant, and more stable traits. The gain matters in clinics, where a model must justify its inputs and keep working as data shifts. The frame targets patient forecast tasks where trust and parsimony count as much as raw score.
Catalog record
| CREATOR | Yuheng Luo, Shuyan Li, Zhong Cao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 15 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-11084