Metraipolis › archive › dc_dc377b739880
Making Clinical Language Models Auditable: Concept-Guided Fine-Tuning for Robust Prediction — arXiv:2608.27397
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.27397
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T14:31:03.255Z
- SHA-256 of the archived text
b590989c0d7967588766087095620ff7850fccffeaa2a41fe6026044f5e6c525- Stable URL of this page
https://metraipolis.com/documento/dc_dc377b739880
How to cite
Making Clinical Language Models Auditable: Concept-Guided Fine-Tuning for Robust Prediction — arXiv:2608.27397 — arXiv (en), https://arxiv.org/abs/2608.27397 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dc377b739880 (Metraipolis archive dc_dc377b739880, sha256 b590989c0d7967588766087095620ff7850fccffeaa2a41fe6026044f5e6c525)
BibTeX
@misc{dcdc377b739880,
title = {Making Clinical Language Models Auditable: Concept-Guided Fine-Tuning for Robust Prediction — arXiv:2608.27397},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.27397},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dc377b739880; sha256 b590989c0d7967588766087095620ff7850fccffeaa2a41fe6026044f5e6c525},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.27397. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Clinical language models can achieve strong in-hospital accuracy yet fail under deployment shifts because they exploit note-specific artifacts (e.g., templates, separators, boilerplate) that do not reflect patient state. We propose CAST (Concept-guided Artifact Suppression Tuning), an SAE-based framework for auditable clinical text classification. CAST uses Sparse Autoencoders to expose sparse, human-auditable features from intermediate Transformer activations, labels SAE latents with an LLM-assisted interpretation pipeline and ICD-10 retrieval constraints, suppresses verified artifact latents via residual subtraction during fine-tuning, and provides post-hoc per-concept attributions for auditing model decisions. On MIMIC-IV discharge-note mortality prediction, CAST improves over its corresponding fine-tuned encoder baselines and remains competitive with strong LLM baselines, while producing a feature-level audit trail of the clinical concepts that support each prediction and the artifact concepts suppressed during training. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-28 in cs.AI). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.27397 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-17T13:16:53.719Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).