Index

Methodology

The Index is designed to be a reviewed professional reference, not a directory scraped from marketing pages.

Inclusion
Verifiable public signal
Review
Primary-source backed
Staleness
Stale after 180 days

Review flow

From inclusion to recheck.

Each record should show why it belongs, which source supports it, what the clinical interpretation is, and when that judgment was last reviewed.

  1. 01 · Inclusion criteria

    Start with a neurosurgical decision point.

    A tool is eligible when it acts directly in neurosurgical care or materially shapes an imaging, planning, monitoring, neuro-oncology, or acute workflow that reaches a neurosurgical team. A generic AI product with no defined clinical task is excluded.

  2. 02 · Primary sources

    Use a source hierarchy.

    Regulatory databases and decision summaries come first, followed by peer-reviewed publications and trial registries. Company material may clarify product features or adoption, but it cannot independently prove regulatory status, clinical validity, or outcome benefit.

  3. 03 · Evidence extraction

    Record what was tested.

    Each row extracts the intended use, algorithm role, dataset or test design, reported performance, human-review requirement, excluded populations, and the strongest conclusion the public evidence supports. Technical validation is not relabeled as clinical outcome evidence.

  4. 04 · Interpretation

    Write the boundary beside the claim.

    Clearance, authorization, and CE marking are not endorsements and do not by themselves show superior outcomes. Every record therefore pairs its evidence signal with a clinical boundary: what the tool does not replace and what remains unproven.

Verification routes

Not every public claim has the same weight.

The Index labels the strongest route that supports a statement so a regulator-confirmed status cannot be confused with a manufacturer report.

Regulator confirmed

Decision letter, public summary, registration record, safety notice, or official regulator list.

Research supported

Peer-reviewed study, protocol, registry entry, or independently accessible validation report.

Manufacturer reported

Official product or company disclosure used only when its origin and evidentiary limit are explicit.

Maintenance

A fixed cadence, plus event-driven updates.

Dates are part of the evidence. A record without a recent review date should not look equally current beside a newly checked one.

  1. Monthly / quarterly

    Regulatory and product status

    Plan monthly safety-notice scans and quarterly checks of clearance, indications, naming and manufacturer. These are editorial targets, not continuous or automated surveillance.

  2. Every 6 months

    Evidence and adoption

    Plan a broader review of external validation, prospective studies and workflow evidence every six months. Match the tested product, version and population; technical performance is not patient benefit.

  3. Event driven

    Major changes

    Update earlier for a new clearance, indication change, withdrawal, safety communication, acquisition, or product retirement.

  4. At 180 days

    Visible stale status

    At editorial review, flag records beyond 180 days without a full recheck. A narrow safety scan or layout edit does not reset the evidence-review date. Keep missed checks visible rather than silently advancing dates.

What the September 6 source check means

The eight original FDA decision records and cited summary sections were rechecked. Product-name searches and device-linked openFDA recall queries identified a scoped SafeOp 3 recall and a later Acorn clearance. Those events predate the August 17–September 6 search window; they are corrections to coverage, not newly occurring events.

The original summary-review dates remain August 17, 2026. September 6 records a targeted source check, not independent clinical peer review or a systematic literature review. Regulatory decision dates remain attached to their individual dossiers.

The retrieved openFDA recall dataset was last updated September 3. No match does not mean no recall or no risk. This check did not exhaustively review MAUDE reports, all jurisdictions, or all product aliases. No product-matched patient-outcome study in the recent window was verified; indexing and access limitations remain.