# Evaluating Large Language Model Raters for German Open-Response Clinical Questions: A Physician-Annotated Benchmark Study of Agreement, Evaluator Bias, and Abstention

arXiv: https://arxiv.org/abs/2607.01103 · HTML: https://mlsanity.com/paper/2607.01103/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 60 references that are in the map.
- First posted on arXiv 1 Jul 2026.

## Builds on (most cited first, 15 of 15)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [What Disease does this Patient Have? A Large-scale Open Domain Question Answering Dataset from Medical Exams](https://mlsanity.com/paper/2009.13081/) (2020)
- [PubMedQA: A Dataset for Biomedical Research Question Answering](https://mlsanity.com/paper/1909.06146/) (2019)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [LLM Evaluators Recognize and Favor Their Own Generations](https://mlsanity.com/paper/2404.13076/) (2024)
- [Prometheus: Inducing Fine-grained Evaluation Capability in Language Models](https://mlsanity.com/paper/2310.08491/) (2023)
- [MedMCQA : A Large-scale Multi-Subject Multi-Choice Dataset for Medical domain Question Answering](https://mlsanity.com/paper/2203.14371/) (2022)
- [Better Zero-Shot Reasoning with Role-Play Prompting](https://mlsanity.com/paper/2308.07702/) (2023)
- [Large Language Models Understand and Can be Enhanced by Emotional Stimuli](https://mlsanity.com/paper/2307.11760/) (2023)
- [LLMEval-Med: A Real-world Clinical Benchmark for Medical LLMs with Physician Validation](https://mlsanity.com/paper/2506.04078/) (2025)
- [Inspection and Control of Self-Generated-Text Recognition Ability in Llama3-8b-Instruct](https://mlsanity.com/paper/2410.02064/) (2024)
- [MORQA: Benchmarking Evaluation Metrics for Medical Open-Ended Question Answering](https://mlsanity.com/paper/2509.12405/) (2025)
- [MediQAl: A French Medical Question Answering Dataset for Knowledge and Reasoning Evaluation](https://mlsanity.com/paper/2507.20917/) (2025)
- [Can Continual Pre-training Bridge the Performance Gap between General-purpose and Specialized Language Models in the Medical Domain?](https://mlsanity.com/paper/2604.19394/) (2026)

## Built on by (most cited first, 2 of 2)

- [Evaluating medical AI under missing information: same-provider judges and human raters change apparent safety](https://mlsanity.com/paper/2607.18828/) (2026)
- [A Shared Taste for Model-Written Text: The Generator-by-Selector Matrices of "AI-AI Bias" Show No Detectable Own-Model Premium](https://mlsanity.com/paper/2610.00369/) (2026)

## Abstract (arXiv)

Background: Expert-annotated benchmarks for non-English open-response clinical questions are scarce. LLM-as-a-judge systems may scale evaluation but require validation. Objective: To introduce MedQADE, a standardized German open-response clinical benchmark with physician reference annotations, and evaluate LLM-as-a-judge alignment, self- and intra-family bias, and abstention. Methods: The benchmark contains 3,800 question-answer sets with answers from five student LLMs and annotations from 10 physicians. All 10 rated the 200-question core; two primary raters assessed each of 3,600 extension questions, with the tenth resolving disagreements. Nine LLM evaluators assessed all sets. We assessed physician reliability, student-model accuracy, evaluator alignment, bias, and abstention. Results: Physicians showed moderate-to-substantial agreement on answer correctness (unweighted mean pairwise Cohen's kappa = 0.612) but limited agreement on question difficulty (Krippendorff's alpha = 0.208 using squared numeric-score distances). Student-model accuracy was 17.8%-66.0% and generally decreased with physician-rated difficulty. Gemini 3 Flash approached the leave-one-out physician reference (kappa = 0.694 vs 0.709). Four of five models rated their own responses more favorably than out-of-family evaluators; five of six intra-family comparisons were positive. Physician abstention increased with perceived difficulty. Seven of nine LLM evaluators abstained in no more than 0.51% of evaluations; the two strongest evaluators assigned definitive labels to every response. Conclusions: Strong LLM evaluators approached physician agreement, but evaluator bias and low observed abstention warrant physician validation and further assessment of selective deferral before fully automated evaluation. These results do not establish clinical safety.