# Medical Bench

> Independent analysis of VQA-RAD and SLAKE: original dataset counts, question variants, image splits, visual grounding, knowledge access and paper-reported results.

A medical VQA result depends on which images, question forms and answer rules were evaluated. We examine VQA-RAD’s clinician-generated questions and SLAKE’s visual and knowledge-based tasks, keeping paper counts separate from cleaned releases. Our original coverage map and guides explain what those choices support. Selected historical results are credited to the original papers; no new model runs are claimed.

## Provenance
Independent analysis published by Arcophos. Benchmark creation belongs to the credited authors. Result rows are selected paper-reported measurements with their source versions and evaluation conditions, not new Arcophos runs or a live leaderboard.

## Benchmark dossiers
- [VQA-RAD](https://medicalbench.ai/benchmarks/vqa-rad/): Count images, question records and question variants separately. Source version: Original 2018 paper and archive.
- [SLAKE](https://medicalbench.ai/benchmarks/slake/): A visual answer and a knowledge answer test different things. Source version: 2021 paper; cleaned SLAKE 1.0 tracked separately.

## Original analyses
- [VQA-RAD: reconcile 3,515 questions with 2,248 records](https://medicalbench.ai/guides/vqa-rad-counts-and-splits/): An original accounting of question forms, image reuse and the original VQA-RAD evaluation.
- [SLAKE: freeze the release, language and knowledge condition](https://medicalbench.ai/guides/slake-version-language-knowledge/): Understand the original image split and why cleaned releases require their own denominators.
- [What VQA-RAD and SLAKE can tell you about visual grounding](https://medicalbench.ai/guides/medical-vqa-visual-grounding/): A task-specific audit of images, answer shortcuts and scoring rules in two medical VQA benchmarks.

## Inspect the evidence
- [Evidence JSON](https://medicalbench.ai/evidence.json): Task definitions, dataset facts, scoring rules, source-version results, our interpretations, and reference IDs.
- [Sources](https://medicalbench.ai/sources/): Original papers and repositories with evidence locators.
- [Editorial method](https://medicalbench.ai/methodology/): Source reconciliation and interpretation boundaries.
- [About](https://medicalbench.ai/about/): Ownership and corrections.

Analysis updated: 2026-09-28