Understanding Structured Health Data through Interaction-Aware Mixture-of-Experts
arXiv:2607.12255
summary
The paper proposes an interaction-aware mixture-of-experts model to predict post-stroke rigidity using multi-level views of structured health records, emphasizing interpretability through routing attribution analysis.
Abstract
We study interaction-aware mixture-of-experts for post-stroke rigidity prediction using multi-level views of structured health records. Despite minimal performance gains, routing attribution reveals systematic importance differences across views, underscoring view construction as key to interpretability.
Topics & keywords
#health data#mixture of experts#post-stroke prediction#interpretability#structured recordsinteraction-aware mixture-of-expertsrouting attributionmulti-level viewsrigidity predictionstructured health records