Evidence dataset for a systematic review of explainable and energy-efficient spiking neural networks for EEG-BCI

This dataset accompanies the systematic review “Explainable and Energy-Efficient Spiking Neural Networks for EEG-BCI: A Systematic Review of Generalisation and Resource-Constrained Deployment.” It contains the definitive source-level evidence ledger used for the review synthesis, with a formal evidence cutoff of 10 August 2026. The master ledger contains 229 source records. Of these, 148 are evidence records comprising 49 primary cognitive/cognitive-affective EEG-SNN records (A1), 58 supporting direct EEG-SNN records (A2), and 41 supporting SNN/neuromorphic/hardware records (B). The direct EEG-SNN evidence subset therefore contains 107 records. The workbook contains source metadata, evidence-boundary classifications, verification notes, generalisation coding, methodological-quality and reporting appraisal (MQRA-v2), temporal summaries, search/corpus provenance, and companion-record decisions. The A1/A2/B evidence categories, G0–G4 generalisation levels, E0–E4 energy-evidence levels, and MQRA-v2 appraisal framework are review-derived analytical structures developed for this review and should not be interpreted as established or externally validated standards. Counts represent publication/evidence records rather than necessarily independent participant cohorts. Historical search provenance is reported separately from later source-level reconstruction and completeness auditing because the original row-level screening/exclusion trail and exact archived literal database search histories are not fully recoverable. Missing historical information has not been reconstructed retrospectively.