# The MedPerturb Dataset: What Non-Content Perturbations Reveal About Human and Clinical LLM Decision Making

arXiv: https://arxiv.org/abs/2506.17163 · HTML: https://mlsanity.com/paper/2506.17163/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Jun 2025.

## Built on by (most cited first, 4 of 4)

- [LiveMedBench: A Contamination-Free Medical Benchmark for LLMs with Automated Rubric Evaluation](https://mlsanity.com/paper/2602.10367/) (2026)
- [MatrAIx: Simulating the World with 8.3 Billion Persona Agents](https://mlsanity.com/paper/2608.04205/) (2026)
- [Compared to What? Baselines and Metrics for Counterfactual Prompting](https://mlsanity.com/paper/2605.01048/) (2026)
- [Chain of Risk: Safety Failures in Large Reasoning Models and Mitigation via Adaptive Multi-Principle Steering](https://mlsanity.com/paper/2605.05678/) (2026)

## Abstract (arXiv)

Clinical robustness is critical to the safe deployment of medical Large Language Models (LLMs), but key questions remain about how LLMs and humans may differ in response to the real-world variability typified by clinical settings. To address this, we introduce MedPerturb, a dataset designed to systematically evaluate medical LLMs under controlled perturbations of clinical input. MedPerturb consists of clinical vignettes spanning a range of pathologies, each transformed along three axes: (1) gender modifications (e.g., gender-swapping or gender-removal); (2) style variation (e.g., uncertain phrasing or colloquial tone); and (3) format changes (e.g., LLM-generated multi-turn conversations or summaries). With MedPerturb, we release a dataset of 800 clinical contexts grounded in realistic input variability, outputs from four LLMs, and three human expert reads per clinical context. We use MedPerturb in two case studies to reveal how shifts in gender identity cues, language style, or format reflect diverging treatment selections between humans and LLMs. We find that LLMs are more sensitive to gender and style perturbations while human annotators are more sensitive to LLM-generated format perturbations such as clinical summaries. Our results highlight the need for evaluation frameworks that go beyond static benchmarks to assess the similarity between human clinician and LLM decisions under the variability characteristic of clinical settings.