Reverse Probing: Supervised Token-level Uncertainty Quantification for Large Language Models in Clinical Text
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913168411852800 |
|---|---|
| author | Xiao, Bushi Soni, Sarvesh Wang, Daisy Zhe |
| author_facet | Xiao, Bushi Soni, Sarvesh Wang, Daisy Zhe |
| contents | As large language models are increasingly deployed for clinical text, ensuring they can reliably signal their own uncertainty becomes critical. Most existing uncertainty quantification (UQ) methods are designed for open-domain generation and cannot localize uncertainty at the token or span level in long clinical text. We propose Reverse Probing, the first UQ framework specialized for clinical summarization, which estimates token-level uncertainty directly from pre-existing labeled summaries. Rather than sampling new outputs, Reverse Probing treats the text as a probe into the model's internal state, extracting uncertainty signals from four categories of internal activations. We evaluate on two expert-annotated clinical datasets and outperform eight adapted baselines on all metrics, achieving up to 4 times higher AUPRC while reducing inference time and computational costs. Feature analysis reveals that delta energy and neighborhood context are the most consistent predictors across all models. This study offers interpretable insights into how models internally respond to unsupported clinical content. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_28740 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Reverse Probing: Supervised Token-level Uncertainty Quantification for Large Language Models in Clinical Text Xiao, Bushi Soni, Sarvesh Wang, Daisy Zhe Computation and Language Artificial Intelligence As large language models are increasingly deployed for clinical text, ensuring they can reliably signal their own uncertainty becomes critical. Most existing uncertainty quantification (UQ) methods are designed for open-domain generation and cannot localize uncertainty at the token or span level in long clinical text. We propose Reverse Probing, the first UQ framework specialized for clinical summarization, which estimates token-level uncertainty directly from pre-existing labeled summaries. Rather than sampling new outputs, Reverse Probing treats the text as a probe into the model's internal state, extracting uncertainty signals from four categories of internal activations. We evaluate on two expert-annotated clinical datasets and outperform eight adapted baselines on all metrics, achieving up to 4 times higher AUPRC while reducing inference time and computational costs. Feature analysis reveals that delta energy and neighborhood context are the most consistent predictors across all models. This study offers interpretable insights into how models internally respond to unsupported clinical content. |
| title | Reverse Probing: Supervised Token-level Uncertainty Quantification for Large Language Models in Clinical Text |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2605.28740 |