GRASP: group-Shapley feature selection for patients
This work addresses feature selection challenges for medical prediction, offering improved robustness and interpretability, though it appears incremental as it builds on existing methods like SHAP and regularization.
The paper tackled the problem of feature selection in medical prediction by introducing GRASP, a framework that combines Shapley value attribution with group L21 regularization, resulting in comparable or superior predictive accuracy while identifying fewer, less redundant, and more stable features.
Feature selection remains a major challenge in medical prediction, where existing approaches such as LASSO often lack robustness and interpretability. We introduce GRASP, a novel framework that couples Shapley value driven attribution with group $L_{21}$ regularization to extract compact and non-redundant feature sets. GRASP first distills group level importance scores from a pretrained tree model via SHAP, then enforces structured sparsity through group $L_{21}$ regularized logistic regression, yielding stable and interpretable selections. Extensive comparisons with LASSO, SHAP, and deep learning based methods show that GRASP consistently delivers comparable or superior predictive accuracy, while identifying fewer, less redundant, and more stable features.