# Rethinking Correctness for Uncertainty Estimation in Clinical Prediction with Vision-Language Models

arXiv: https://arxiv.org/abs/2609.15180 · HTML: https://mlsanity.com/paper/2609.15180/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Clinical and biomedical applications, one of 17 directions in LLM reasoning.
- Builds on 18 of its 24 references that are in the map.
- First posted on arXiv 14 Sep 2026.

## Placed in

- LLM reasoning / [Clinical and biomedical applications](https://mlsanity.com/lr/clinical-and-biomedical-applications/)
- Multimodal LLMs / vision-language models / [Medical vision-language models](https://mlsanity.com/mllm/medical-vision-language-models/)

## Builds on (most cited first, 18 of 18)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://mlsanity.com/paper/2006.03654/) (2020)
- [SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models](https://mlsanity.com/paper/2303.08896/) (2023)
- [Humans or LLMs as the Judge? A Study on Judgement Biases](https://mlsanity.com/paper/2402.10669/) (2024)
- [Gemma 4 Technical Report](https://mlsanity.com/paper/2607.02770/) (2026)
- [Scalable Best-of-N Selection for Large Language Models via Self-Certainty](https://mlsanity.com/paper/2502.18581/) (2025)
- [Evaluating Correctness and Faithfulness of Instruction-Following Models for Question Answering](https://mlsanity.com/paper/2307.16877/) (2023)
- [EHRSHOT: An EHR Benchmark for Few-Shot Evaluation of Foundation Models](https://mlsanity.com/paper/2307.02028/) (2023)
- [Tomayto, Tomahto. Beyond Token-level Answer Equivalence for Question Answering Evaluation](https://mlsanity.com/paper/2202.07654/) (2022)
- [MedGemma 1.5 Technical Report](https://mlsanity.com/paper/2604.05081/) (2026)
- [Estimating LLM Uncertainty with Evidence](https://mlsanity.com/paper/2502.00290/) (2025)
- [Revisiting Uncertainty Quantification Evaluation in Language Models: Spurious Interactions with Response Length Bias Results](https://mlsanity.com/paper/2504.13677/) (2025)
- [Enhancing Hallucination Detection through Noise Injection](https://mlsanity.com/paper/2502.03799/) (2025)
- [Addressing Pitfalls in the Evaluation of Uncertainty Estimation Methods for Natural Language Generation](https://mlsanity.com/paper/2510.02279/) (2025)
- [Cross-Representation Benchmarking in Time-Series Electronic Health Records for Clinical Outcome Prediction](https://mlsanity.com/paper/2510.09159/) (2025)
- [The Taxonomies, Training, and Applications of Event Stream Modelling for Electronic Health Records](https://mlsanity.com/paper/2603.14003/) (2026)
- [Towards Generation-Efficient Uncertainty Estimation in Large Language Models](https://mlsanity.com/paper/2605.06053/) (2026)
- [From Token to Token Pair: Efficient Prompt Compression for Large Language Models in Clinical Prediction](https://mlsanity.com/paper/2605.11774/) (2026)

## Abstract (arXiv)

Vision-language models are increasingly explored for clinical prediction from electronic health records and medical images, where identifying unreliable predictions is important for safe deployment. Uncertainty estimation (UE) enables detecting such predictions, but its evaluation depends on a correctness criterion that determines whether each model output is correct. If this criterion disagrees with human judgement or distorts downstream UE performance, conclusions about model reliability can be misleading. We introduce a two-axis framework that evaluates correctness criteria by their agreement with human judgements and fidelity to human-referenced UE performance. We assess eight criteria across three clinical prediction tasks and three models using 450 predictions annotated by two reviewers. Across the audited tasks, canonical exact matching (EM) achieved the highest observed human agreement and lowest UE distortion, while the BERT-based matching (BEM) and LLM-judge also showed strong human agreement. Across four UE methods and 23,254 clinical predictions, criterion choice changed error-detection AUROC by up to 0.146 and reversed the relative ranking of UE methods. The LLM-judge also selectively accepted invalid or uncertain outputs, accepting 16 of 30 such human-identified errors. These results demonstrate that correctness assessment is an integral component of clinical UE evaluation and should be validated before UE methods are compared.