About the project
A community-driven, centralized resource of anonymized clinical headache cases with validated ICHD-3 diagnoses, with a leaderboard comparing how well language models diagnose them.
Clinicians contribute cases, headache specialists review them, and researchers download citable dataset releases to evaluate how well large language models diagnose headache disorders. The platform hosts the resulting benchmark leaderboard; it does not run model evaluations itself.
Neither the cases nor the benchmark results are a diagnostic tool — they are a research resource and must not be used for clinical decision-making.
The paper
The platform realizes the vision set out in:
Anika Zahn, Sebastian Strauss, Dorian Zwanzig — “Towards Community-Based Evaluation of AI in Neurology: Development of a Headache Diagnosis Dataset for Large Language Models”.
Full publication details will be added here once available. For citing the dataset itself, use the citation hint on the public dataset page.
Team
- Anika Zahn — Department of Neurology, University Medicine Greifswald
- Sebastian Strauss — Department of Neurology, University Medicine Greifswald
- Dorian Zwanzig — Eberswalde University for Sustainable Development (HNEE), Eberswalde
Ethics approval
The underlying multi-center study, “KI-unterstützte ICHD-3-Klassifikation primärer Kopfschmerzerkrankungen”, received a favorable opinion from the Ethics Committee of Universitätsmedizin Greifswald (reg. no. BB 211/25, dated 2025-11-26). Study data are handled separately from the public community-tier cases on this site and are never publicly accessible.
Contact
Reach the project team at [email protected].
Concerns about a specific case — for example a re-identification risk — go through the report form, which needs no account.