Data and code for Motor imagery–based brain–computer interfaces for stroke rehabilitation: a bibliometric and visualization analysis

This dataset accompanies a bibliometric and visualization analysis of motor imagery–based brain–computer interfaces for stroke rehabilitation. The corpus comprises 541 publication reports assembled from Web of Science, Scopus, PubMed, China National Knowledge Infrastructure, Wanfang, and VIP in October 2026. The report is the unit of analysis; separate publications from the same clinical study may be represented separately.The files contain report identifiers, publication years, original bibliographic titles, available DOI/PMID identifiers, source-database labels, standardized per-report keyword inputs, a keyword mapping dictionary, aggregate publication and collaboration statistics, keyword and reference co-citation network data, temporal frequency and burst outputs, sensitivity analyses, database search expressions, and portable Python verification and reproduction scripts. The code reproduces the keyword networks, period frequencies, and burst analyses from the supplied standardized inputs. Collaboration and reference co-citation files are aggregate analytical outputs and do not reconstruct the original full author/affiliation lists or cited-reference exports.There are 540 reports with known publication years: 473 during 2008–2025 and 67 in the incomplete year 2026. The primary keyword analysis uses 475 reports with usable keyword information; analyses restricted to complete years use 411 keyword-available reports. Reference co-citation outputs use the 377-report Web of Science subset with complete exported reference counts. Missing information is not imputed as zero.The files do not include article abstracts, full texts, participant-level data, private contact details, credentials, or raw licensed database exports. Original bibliographic titles and lexical source terms are retained for traceability. Deterministic computational outputs are derived from the existing bibliographic corpus; they are not synthetic bibliographic observations generated by a language model. Computational checks establish internal consistency and reproducibility of the supplied analytical inputs, rather than independently establishing eligibility or the accuracy of the source publications.Licensing: CC BY 4.0 applies to the authors' own analytical data and documentation. MIT applies to the original Python scripts. Third-party materials and external libraries retain their own rights. See the license files and third-party notice in the archive.