# Expert Evaluation of Clinical AI Tools on Real Point-of-Care Clinical Queries

arXiv: https://arxiv.org/abs/2606.28960 · HTML: https://mlsanity.com/paper/2606.28960/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 31 references that are in the map.
- First posted on arXiv 27 Jun 2026.

## 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)
- [Large Language Models Encode Clinical Knowledge](https://mlsanity.com/paper/2212.13138/) (2022)
- [What Disease does this Patient Have? A Large-scale Open Domain Question Answering Dataset from Medical Exams](https://mlsanity.com/paper/2009.13081/) (2020)
- [Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference](https://mlsanity.com/paper/2403.04132/) (2024)
- [BERTopic: Neural topic modeling with a class-based TF-IDF procedure](https://mlsanity.com/paper/2203.05794/) (2022)
- [ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate](https://mlsanity.com/paper/2308.07201/) (2023)
- [HealthBench: Evaluating Large Language Models Towards Improved Human Health](https://mlsanity.com/paper/2505.08775/) (2025)
- [Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Models](https://mlsanity.com/paper/2404.18796/) (2024)
- [WildBench: Benchmarking LLMs with Challenging Tasks from Real Users in the Wild](https://mlsanity.com/paper/2406.04770/) (2024)
- [Benchmarking Cognitive Biases in Large Language Models as Evaluators](https://mlsanity.com/paper/2309.17012/) (2023)
- [Adding Error Bars to Evals: A Statistical Approach to Language Model Evaluations](https://mlsanity.com/paper/2411.00640/) (2024)
- [Medical Large Language Model Benchmarks Should Prioritize Construct Validity](https://mlsanity.com/paper/2503.10694/) (2025)
- [Neither Valid nor Reliable? Investigating the Use of LLMs as Judges](https://mlsanity.com/paper/2508.18076/) (2025)
- [Prediction-Powered Ranking of Large Language Models](https://mlsanity.com/paper/2402.17826/) (2024)
- [Human-AI Co-design for Clinical Prediction Models](https://mlsanity.com/paper/2601.09072/) (2026)
- [JudgmentBench: Comparing Rubric and Preference Evaluation for Quality Assessment](https://mlsanity.com/paper/2605.25240/) (2026)
- [LLMs Judging LLMs: A Simplex Perspective](https://mlsanity.com/paper/2505.21972/) (2025)
- [Statistical inference with win statistics in cluster-randomized trials with hierarchical composite outcomes](https://mlsanity.com/paper/2604.18341/) (2026)

## Built on by (most cited first, 5 of 5)

- [Agentic systems for breast cancer treatment recommendations](https://mlsanity.com/paper/2607.12051/) (2026)
- [Judge-dependent safety gains and model-specific helpfulness costs of evidence-sufficiency prompting in clinical LLMs](https://mlsanity.com/paper/2607.18086/) (2026)
- [Scaling Clinical Judgment to Evaluate Medical AI](https://mlsanity.com/paper/2609.12822/) (2026)
- [Verifiable by Construction: Claim-Level Evaluation of Verbatim Citation in Clinical Question Answering](https://mlsanity.com/paper/2609.15964/) (2026)
- [Clinician use of language models diverges from how the models are evaluated](https://mlsanity.com/paper/2610.11069/) (2026)

## Abstract (arXiv)

Physicians now pose millions of clinical questions to AI tools each week, yet these tools are evaluated largely on hypothetical or exam-style questions, not those actually asked in practice. We report a blinded evaluation built on 620 Real-world Point-Of-Care Queries (Real-POCQi) submitted to the OpenEvidence (OE) platform by physicians spanning 30 specialties, as well as 187 questions from HealthBench. 149 practicing physicians across 36 states made head-to-head comparisons between answers from three frontier general-purpose models (Claude Opus 4.8, Gemini 3.1 Pro, and GPT-5.5) and a specialized clinical tool (OE), with graders matched to each question's specialty. When comparing answers along five dimensions relevant to clinical decision support -- accuracy, clinical utility, source quality, verifiability, & completeness -- physicians scored the specialized tool highest on all axes; in the primary analysis on Real-POCQi, win differences (margins between win and loss rates) ranged from 25 to 39 percentage points (p<0.001). Results remained consistent in sensitivity analyses stratifying by citation display, answer length, OE-user status, and Real-POCQi versus HealthBench. In parallel, LLM judges were found to systematically differ from expert judges, though both generally agreed on the best model. These findings underscore two conclusions: (i) AI tool evaluations should reflect real-world query distributions and use expert judges that mirror the specialization defining modern medicine and (ii) the consistent advantage of the specialized tool over general-purpose models does not necessarily mean that the latter cannot serve similar purposes, but that targeted engineering and customization can yield meaningful gains in performance for its users. We release Real-POCQi as a public benchmark, as well as the prespecified statistical analysis for reproducing results of this study.