GlobalDentBench: A Multinational Benchmark for Evaluating LLM Clinical Reasoning in Dentistry with Expert Calibration

2026-05-23

AI Summary

This study addresses the lack of systematic evaluation of robustness and safety in large language models (LLMs) for dental clinical reasoning. The authors introduce the first multinational dental benchmark, encompassing 88 countries across six continents, featuring 8,978 expert-validated questions spanning 14 specialties. They propose a novel three-tiered clinical reasoning difficulty framework—knowledge recall, routine reasoning, and personalized reasoning—and incorporate diverse question formats, including multiple-choice, short-answer, and case-based analyses. Leveraging an expert-calibrated automated construction pipeline, the benchmark achieves 99.98% consistency in question generation. Evaluation of 12 state-of-the-art LLMs reveals alarming limitations: accuracy drops to 22.34% on case-based questions and further declines to 35.71% across the three reasoning tiers, with 31.01% of responses posing safety risks—including 4.51% potentially causing irreversible harm—highlighting significant gaps in current models’ readiness for real-world clinical deployment.

Abstract

While large language models (LLMs) hold transformative potential for medicine, their reasoning robustness and safety in real-world clinical scenarios remain critically underexplored, particularly in dentistry. Here we introduce GlobalDentBench, the first multinational dental benchmark, featuring a taxonomy that encompasses 14 dental specialties across 88 countries and regions spanning six continents. The benchmark comprises 8,978 expert-validated questions across three formats (multiple-choice, short-answer, and case-based questions) and assesses three progressive reasoning levels: knowledge recall (L1), routine reasoning (L2), and individualized reasoning (L3). To ensure data quality, the automated construction framework was calibrated by six senior dentists, achieving expert agreement rates of 99.98% for multiple-choice and short-answer questions and 96.78% for the more complex case-based questions. Evaluation of 12 frontier LLMs on GlobalDentBench revealed a sharp, stepwise performance degradation with increasing reasoning complexity. Specifically, accuracy plummeted from 81.34% on multiple-choice to 64.53% on short-answer and 22.34% on case-based questions, while declining markedly from 74.01% at L1 to 55.64% at L2 and 35.71% at L3. More critically, risk analysis of real-world dental cases demonstrated an alarming overall unsafe rate of 31.01% in LLM-generated clinical recommendations, with 4.51% posing risks of irreversible patient harm and risks particularly pronounced in specialties such as orthodontics. These findings expose fundamental limitations in the medical reasoning and safety of current LLMs. Consequently, GlobalDentBench provides a scalable foundation for trustworthy clinical AI evaluation, underscoring the urgent need for rigorous validation before the safe deployment of these models in healthcare.

Problem

• large language models
• clinical reasoning
• dental AI safety
• multinational benchmark
• LLM evaluation

Innovation

• GlobalDentBench
• clinical reasoning evaluation
• expert-calibrated benchmark
• LLM safety in dentistry
• multinational dental dataset