Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2604.01841 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913177418072064 |
|---|---|
| author | Pham, Minh-Khoi Ho, Thang-Long Nguyen Dao, Thao Thi Phuong Mai, Tai Tan Tran, Minh-Triet Ward, Marie E. Geary, Una Brennan, Rob McDonald, Nick Crane, Martin Bezbradica, Marija |
| author_facet | Pham, Minh-Khoi Ho, Thang-Long Nguyen Dao, Thao Thi Phuong Mai, Tai Tan Tran, Minh-Triet Ward, Marie E. Geary, Una Brennan, Rob McDonald, Nick Crane, Martin Bezbradica, Marija |
| contents | Clinical prediction from structured electronic health records (EHRs) is challenging due to high dimensionality, heterogeneity, class imbalance, and distribution shift. While tabular in-context learning (TICL) and retrieval-augmented methods perform well on generic benchmarks, their behavior in clinical settings remains unclear. We present a multi-cohort EHR benchmark comparing classical, deep tabular, and TICL models across varying data scale, feature dimensionality, outcome rarity, and cross-cohort generalization. PFN-based TICL models are sample-efficient in low-data regimes but degrade under naive distance-based retrieval as heterogeneity and imbalance increase. We propose AWARE, a task-aligned retrieval framework using supervised embedding learning and lightweight adapters. AWARE improves AUPRC by up to 12.2% under extreme imbalance, with gains increasing with data complexity. Our results identify retrieval quality and retrieval-inference alignment as key bottlenecks for deploying tabular in-context learning in clinical prediction. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_01841 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Retrieval-aligned Tabular Foundation Models Enable Robust Clinical Risk Prediction in Electronic Health Records Under Real-world Constraints Pham, Minh-Khoi Ho, Thang-Long Nguyen Dao, Thao Thi Phuong Mai, Tai Tan Tran, Minh-Triet Ward, Marie E. Geary, Una Brennan, Rob McDonald, Nick Crane, Martin Bezbradica, Marija Artificial Intelligence Clinical prediction from structured electronic health records (EHRs) is challenging due to high dimensionality, heterogeneity, class imbalance, and distribution shift. While tabular in-context learning (TICL) and retrieval-augmented methods perform well on generic benchmarks, their behavior in clinical settings remains unclear. We present a multi-cohort EHR benchmark comparing classical, deep tabular, and TICL models across varying data scale, feature dimensionality, outcome rarity, and cross-cohort generalization. PFN-based TICL models are sample-efficient in low-data regimes but degrade under naive distance-based retrieval as heterogeneity and imbalance increase. We propose AWARE, a task-aligned retrieval framework using supervised embedding learning and lightweight adapters. AWARE improves AUPRC by up to 12.2% under extreme imbalance, with gains increasing with data complexity. Our results identify retrieval quality and retrieval-inference alignment as key bottlenecks for deploying tabular in-context learning in clinical prediction. |
| title | Retrieval-aligned Tabular Foundation Models Enable Robust Clinical Risk Prediction in Electronic Health Records Under Real-world Constraints |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2604.01841 |