MoCA-Dialog: A Benchmark for Fine-Grained Evaluation of Large Language Models in Clinical Cognitive Assessment Dialogues
Bibliographic record
Abstract
Current benchmarks for evaluating large language models (LLMs) in medicine primarily focus on static questionanswering, neglecting the dynamic, interactive nature of many clinical workflows. This is particularly evident in cognitive assessments, where nuanced dialogue and multimodal interpretation are critical, yet no specialized benchmark exists to evaluate these capabilities. To address this gap, we introduce MoCADialog, the first large-scale, multimodal benchmark featuring 5000 high-fidelity simulated dialogues for the Montreal Cognitive Assessment (MoCA). The benchmark includes tasks of increasing complexity: accurate scoring, cognitive profile generation, and clinical error attribution, allowing for a fine-grained evaluation across seven cognitive domains. Our comprehensive evaluation of state-of-the-art multimodal LLMs reveals significant performance disparities; while models excel at simple recall tasks, they consistently fail in domains requiring executive function and abstract reasoning. A novel, clinically-driven error analysis further indicates that these failures stem not from knowledge deficits, but from fundamental difficulties in interpreting nuanced cues and applying domain-specific reasoning. MoCA-Dialog provides a crucial tool for assessing the clinical readiness of LLMs and highlights that future progress depends on enhancing their core reasoning and interpretive abilities, not just expanding their knowledge base. We release our demo, and prompt examples at https://mocadialogue.github.io.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.022 | 0.109 |
| Meta-epidemiology (narrow) | 0.004 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.005 | 0.004 |
| Open science | 0.005 | 0.008 |
| Research integrity | 0.004 | 0.003 |
| Insufficient payload (model declined to judge) | 0.009 | 0.005 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".