BCI Report Explorer

by twu31co.huggingface.twu31/bci-report-explorer

  • Remote endpoint
  • 7 tools
  • MIT

Tools

  • bci_report_explorer_cite How to cite BCI Report, its licences, links, and what this Space indexes. Tool names: Gradio API /cite; MCP on the Hugging Face Space bci_report_explorer_cite. Returns: JSON text: the Zenodo concept DOI and BibTeX, the licences of the aggregate results (CC BY 4.0), the code (MIT) and the recordings (their own), and project links.
  • bci_report_explorer_get_dataset_results Every indexed figure computed on one dataset, with its licence and credit. Not a ranking: compare only within one protocol or group. Tool names: Gradio API /get_dataset_results; MCP on the Hugging Face Space bci_report_explorer_get_dataset_results. Returns: JSON text: dataset licence and credit lines, and rows grouped by comparison group. An unknown dataset is a tool error listing candidates.
  • bci_report_explorer_get_method_results Every indexed figure for one method, grouped by protocol or topic. Not a ranking: compare only within one protocol or group. Paired contrasts between two methods are listed under both. Matching is case-insensitive; a unique partial name also works. Tool names: Gradio API /get_method_results; MCP on the Hugging Face Space bci_report_explorer_get_method_results. Returns: JSON text: rows grouped by comparison group, each with value, interval, n, chance and caveats; fields shared by a group are given once on the group. An unknown name is a tool error listing candidates.
  • bci_report_explorer_get_protocol_results Every published method row for one core protocol, in published order. Values are copied unchanged from the pinned BCI Report dataset. Rows are not ranked; compare them only within this protocol. EEGPT, frozen, follows in its own block (eegpt_frozen), beside the core matrix: grouped, never ranked, called above or below a published row only where the file says their 95% intervals do not overlap, with the gap. Tool names: Gradio API /get_protocol_results; MCP on the Hugging Face Space bci_report_explorer_get_protocol_results. Returns: JSON text: the protocol descriptor and its rows (value, 95% interval, n, chance, caveats); release_id, source_url and other fields shared by all rows are given once beside the rows.
  • bci_report_explorer_get_question One of the site's seventeen questions: its short answer, paired contrasts and measurements. Tool names: Gradio API /get_question; MCP on the Hugging Face Space bci_report_explorer_get_question. Returns: JSON text: the question, the short answer as the site publishes it, paired contrasts with intervals and verdict wording as given, measurements, and related held items (for questions-in-language also the wordings its page shows; for new-ssvep-frequencies what each card did not run; for data-quality its two status cases). The pilot-studies, beyond-scalp-eeg, clinical-groups, data-quality and new-ssvep-frequencies answers carry their notices: pilots and NeuroLM's panel sit outside every leaderboard, rights gaps, EEGPT's licence status, the clinical disclaimer, MC_RTT's one animal and ds007420's counts-only check.
  • bci_report_explorer_list_holds Results that are held, status-only, withheld, not run or not published. Listed by name, status and reason only. No figure is given for any of them, because none is published. Tool names: Gradio API /list_holds; MCP on the Hugging Face Space bci_report_explorer_list_holds. Returns: JSON text: held items grouped by source file and release (each with its name, status, reason and source URL), and, on request, the not-published register by release.
  • bci_report_explorer_list_protocols List the eight fixed core protocols of BCI Report (no scores). Each protocol is a separate comparison with its own cohort, metric and chance level, so figures are comparable only within one protocol. Use a returned protocol_id with get_protocol_results. Tool names: Gradio API /list_protocols; MCP on the Hugging Face Space bci_report_explorer_list_protocols. Returns: JSON text: protocols with id, title, dataset, cohort size, metric, chance level, limitation and page URL, plus provenance and notices.

As the server's own listing gives them. We never call a tool.