FILE 03 / EVIDENCE AUDIT
The source map is being checked before it is republished.
Citation verification in progress. The earlier downloadable review remains withdrawn. Since the previous revision the project has recorded a partial claim-by-claim audit, demoted its research briefs to non-controlled evidence notes, and written a controlled specification that keeps requirements separate from literature.
No citation count or quantitative literature summary on this page should be treated as validated until the source-by-source audit is complete.
Withdrawn
Excluded from the controlled source set.
None
No prototype or participant dataset.
Partial
Superseded as the decision record.
One
Public EEG data, not a device result.
01 / WHY WITHDRAW
A working link is not proof that a citation supports a claim.
Release review found references whose DOI or arXiv identifier resolved to an unrelated work. Because those references supported biomedical and engineering statements, correcting only the bibliography would not be enough.
The paper will remain unavailable until every public claim has been traced to a source that was read and verified. This is a content-integrity correction, not evidence that Norysr has failed or succeeded technically.
02 / WHAT THE AUDIT HAS FOUND SO FAR
Most high-impact identifiers resolved. Some did not, and the tally itself is not yet reproducible.
A first-pass audit (August 2026) traced the high-impact numerical claims in the project's five research briefs to their identifiers through registry lookups and tagged each claim as verified, unverified, disputed or hypothesis. Most clinical-efficacy and inertial-sensing identifiers resolved to the attributed works with matching sample sizes and results.
It also found real defects. One brief attached a pediatric cerebral-palsy gait-event result to an identifier that belongs to a different author group, conflating two papers. Several preprint identifiers could not be resolved during the audit window. And because the audit was reconstructed after an interruption, its headline totals cannot be reproduced from the rows it lists, so the project treats it as evidence work, not as a decision record.
The consequence is written into the controlled specification: the earlier stimulation, ramp, burst, refractory, charge-cap and gait-window values are explicitly not adopted as requirements, literature values are never promoted to product requirements automatically, and the research briefs are labelled non-controlled evidence notes.
03 / AUDIT METHOD
Four checks must pass before evidence returns.
Match identifiers
Resolve each DOI, PMID and arXiv identifier against an authoritative metadata record.
Read the source
Check that the paper actually reports the population, task, hardware and result attributed to it.
Trace every number
Connect each quantitative statement to a verified source or remove it from the public dossier.
Separate proposal from evidence
Keep Norysr targets, budgets and regulatory assumptions visibly distinct from published findings.
04 / DATA CANDIDATES
No dataset in the workspace is ready for a gait-model claim.
Pediatric gait candidate
A public 2026 multi-sensor pediatric gait deposit with cerebral-palsy, toe-walking and typically developing children was reviewed as a development-tier candidate only: its licence and derivative rights are unconfirmed, its lower-leg sensor placement differs from the proposed distal shank, and its released gait events derive from the same angular-velocity signal the model would consume, so it cannot validate anything.
EEG emotion table
A two-person EEG feature table found in the workspace holds summary statistics, not waveforms, and no participant identifiers. It is ineligible for every Norysr tier and was used only in the side study, where a linear projection out-performed a neural network at restoring its features.
Public motor EEG
A public 109-person motor-movement EEG dataset under an open attribution licence supplied the side study’s 24 subjects; every file was hash-verified against the publisher’s checksums and recorded in a manifest.
Required next
A participant-grouped dataset with pediatric CP, a distal-shank inertial sensor and an independent synchronized gait-event reference, collected or licensed under an approved protocol. Nothing in the workspace satisfies this.
05 / CURRENT BOUNDARY
The engineering proposal remains unvalidated.
- No completed Norysr wearable prototype exists and no hardware has been selected.
- No participant has been enrolled and no clinical outcome has been measured.
- No latency, accuracy, energy or stimulation result has been measured on Norysr hardware, and no gait model has been trained.
- No regulatory classification or predicate strategy has been confirmed for Norysr.
- No hospital, university, clinician or manufacturer partnership is represented.
- The EEG side study is a computational result on public healthy-adult data and does not change any statement above.