@inproceedings{jalan-etal-2026-medbench,
    title = "{M}ed{B}ench: Deliberative Evaluation of Medical Language Models",
    author = "Jalan, Pratik  and
      Joshi, Mukul  and
      Magotra, Akhilesh  and
      Jadhav, Kshitij",
    editor = "Demner-Fushman, Dina  and
      Ananiadou, Sophia  and
      Roberts, Kirk  and
      Tsujii, Junichi",
    booktitle = "{B}io{NLP} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.bionlp-1.79/",
    pages = "981--991",
    ISBN = "979-8-89176-434-7",
    abstract = "We introduce MedBench, a benchmark for evaluating medical language models as deliberating agents rather than isolated predictors. MedBench evaluates eight models (4B?32B) on 19,625 questions from six medical QA datasets using Consensus-Aware Model Panel (CAMP), a two-tier protocol in which five 4B?8B models answer independently, revise after observing peer reasoning, and escalate persistent disagreements to larger 20B?32B models. Compared with zero-shot, few-shot, and chain-of-thought baselines, CAMP shows that deliberation is not uniformly accuracy-improving, but reveals interaction-driven behaviors hidden by single-model evaluation. On PubMedQA without external context, the 4B?8B panel outperforms the evaluated 20B?32B individual zero-shot models (54.1{\%} vs. 33.9{\%}), and achieves the best evaluated result with context (75.7{\%}), suggesting that structured interaction can sometimes complement scale. Across five datasets, initial inter-model agreement is positively associated with correctness and serves as a useful difficulty signal. However, on MedXpertQA, unanimous agreement yields only 6.6{\%} accuracy despite 14.4{\%} overall accuracy, suggesting correlated ignorance, where shared biases make consensus misleading. Error analysis shows that most failures are debate-insufficient cases, where incorrect majorities persist despite interaction (93?97{\%}), while debate-harmful cases account for 3?7{\%}. MedBench positions deliberative evaluation as a complement to accuracy-centric benchmarking, measuring when model interaction corrects errors, reinforces shared mistakes, or signals the need for stronger evidence and human review."
}