See how VitaSignal scopes a clinical AI decision, evaluates the local documentation pathway, records evidence and limitations, applies risk controls, and prepares a Governance Decision Package.
Tour data: generated, no PHI.VitaSignal recommends; the health system decides. Tour records are generated examples and are not part of the empirical evidence record. No customer outcome is shown. Findings for your setting come from a scoped review of your own retrospective or approved read-only data. Detailed limitations are documented on the Evidence and Trust pages.
Everything the interactive tour covers, in reading order, for anyone who prefers text or cannot use the frame above.
VitaSignal conducts a Clinical AI Documentation Readiness Evaluation for health-system governance committees, informatics teams, and executive reviewers weighing one proposed clinical AI use case.
The tour presents two supporting views of that single evaluation, a Governance Review Workspace and a Data Pathway Readiness Workspace. They are perspectives on the same work, not separate products.
Each path states the intended use, the permitted use, the prohibited use, the accountable decision owner, and the evidence threshold that must be met before anything advances.
Reviews are retrospective or run through an approved read-only pathway. There is no chart write-back, no patient-care direction, and no clinical use.
Timing and cutoff, missingness, mapping gaps, rejected records, provenance, and source readiness are reviewed and recorded before any result is interpreted.
Any real evaluation requires an approved least-privilege, read-only pathway; this tour uses generated records and makes no live connection.
The Local Readiness Review leads as the first and default scenario. It reports added discrimination observed in an institutionally external confirmation, and it keeps the adverse findings in the same view: a worse Brier score, a calibration slope well below one, low positive predictive value, degradation with missingness, and unavailable demographic subgroup, hospital, and vendor analyses. The record is an institutionally external retrospective, historical and not historically untouched.
A favorable discrimination direction does not erase those limits. No clinical utility, patient benefit, generalizability, deployment readiness, or complete-product validation is established.
A separate endpoint-feasibility example is shown as blocked: the required extubation and 48-hour failure endpoint could not be reconstructed defensibly, so no proxy endpoint was substituted and no model was run. It is neither a success nor a failure of the Local Readiness Review.
Generated tour records are kept visibly separate from referenced retrospective results, and no customer outcome is shown.
Feature-parity and hospital-aware transfer gates did not close, so the workflow stopped before scoring. No model was scored, no target-label recalibration occurred, and no performance metric was produced.
That case is presented as a Stop. It is evidence that the evaluation halts when required conditions are absent, not evidence of successful transfer or clinical benefit.
Every material risk is paired with a required control, a named accountable owner, an explicit stopping condition, and the specific next evidence that would be needed.
Unresolved questions stay visible rather than being closed out. The final institutional decision remains a human one.
The tour ends in one versioned record covering scope, source provenance, evidence state, findings marked favorable, adverse, blocked, absent, or uncertain, risks, controls, owners, unresolved questions, next evidence, and a recommendation of Proceed, Modify, Pause, or Stop.
Study Not Justified is one of the reasons a recommendation can be Stop.
The tour includes a context mapping of published guidance on AI lifecycle governance, risk-proportionate review, governance readiness, transparency, equity and monitoring, the difference between model type and regulatory or evidence status, human decision rights, shared regional governance, independent generator-and-reviewer verification, and standards-version tracking.
Items are labelled Adopted, Mapped, Conditional, Not adopted, or Unresolved only where that judgement is defensible. Nothing there implies endorsement, compliance, certification, or product validation, and any item without a verified citation on this site is labelled a contextual theme requiring specific source review.
VitaSignal recommends; the health system decides.