Vertical pack — Clinical & Health

    Clinical & Health AI Assurance.

    An independent assurance overlay above clinical decision support, triage, documentation and administrative health AI. We do not adjudicate clinical judgement. We evidence that the model was authorised for the task, that patient data stayed inside its lawful boundary, that a named clinician owned the decision, and that the episode can be reconstructed on request.

    Cabier does not practise medicine, does not issue clinical guidance and does not replace the organisation's own clinical governance committee.

    The obligation set

    Resolved into testable control statements. Applicability is confirmed per engagement — jurisdiction, product and deployment scope all change what binds.

    HIPAA Privacy & Security Rules

    Minimum necessary, access control and audit-log expectations applied to prompts, retrieved records and model outputs — not only to the source EHR.

    Clinical governance

    Local approval of the tool for a defined clinical task, at a defined version, with a named clinical owner and a defined escalation path.

    Medical-device software expectations

    Where an output influences diagnosis or treatment, evidence of intended-use scope, change control and post-deployment monitoring.

    EU AI Act high-risk duties

    For in-scope deployments: human oversight, logging, accuracy and robustness evidence held by the deployer, not asserted by the vendor.

    Non-discrimination and equity duties

    Subgroup performance tested and recorded where the model influences access, triage priority or resource allocation.

    Pack modules

    Each module is a vertical projection of an asset already running on the platform. No parallel store, no second evidence chain.

    Clinical AI Evidence Vault

    Unified Evidence Vault

    Prompt, retrieved patient context references, output, reviewing clinician and disposition captured independently of the vendor console, with an audit identifier per episode.

    PHI Boundary Control

    Sovereign AI Governance OS — residency and PII gating

    Tested prevention of protected health information reaching an unapproved model, region or retention path — with the block recorded, not silently dropped.

    Clinical Task Registry

    AI Model Inventory

    Which model is approved for which clinical or administrative task, at which version, under whose clinical authority, and when that approval lapses.

    Clinician Accountability Log

    Unified Audit Trail

    An ordered record of who invoked the model, what it returned, which clinician accepted, modified or rejected it, and what entered the record.

    What the institution gets that a horizontal tool cannot give it

    Clinical AI Governance Maturity Score

    A scored position across approval, PHI boundary, logging, clinician accountability, subgroup performance, patient transparency, drift monitoring and reconstruction — refreshed continuously.

    PHI Containment Score

    A separate measure of whether protected health information can leave its lawful boundary at all, tested against live paths rather than asserted in policy.

    Clinical Exposure View

    Where unreviewed, unlogged or out-of-scope model use is accumulating by service line, surfaced before it becomes an incident or a claim.

    Board & Regulator Attestation Pack

    A scoped attestation for the clinical governance committee, payer or supervisor, naming period, systems in scope and stated limitations.

    Clinical AI governance maturity — self-assessment

    Eight questions across the dimensions a clinical incident review or regulatory inspection would actually test. Answers stay in your browser — nothing is submitted, stored or transmitted.

    01 · Approval

    Is every AI tool touching patient care approved for a defined clinical task, at a defined version, with a named clinical owner?

    02 · PHI boundary

    Is there a tested control preventing protected health information reaching an unapproved model, vendor or region?

    03 · Logging

    Are prompts, retrieved context and outputs retained independently of the vendor's own console?

    04 · Accountability

    Is a named clinician recorded against every AI-influenced entry that reaches the patient record?

    05 · Subgroup performance

    Has performance been tested across relevant patient subgroups and the result recorded?

    06 · Patient transparency

    Is there a defined position on whether and how patients are told AI contributed to their care?

    07 · Drift monitoring

    Is post-deployment performance monitored on your own population rather than the vendor's validation set?

    08 · Reconstruction

    Could you reconstruct, on request, how a specific AI-influenced clinical decision was produced?

    Answer the questions above to see an indicative position.

    The 90-day pilot

    Days 1–15

    Scope and obligation mapping

    Service lines in scope, tools in use, obligation set fixed, evidence types and retention agreed with clinical governance.

    Days 16–45

    Capture and PHI boundary

    Independent evidence capture stood up; PHI boundary and residency controls tested against live paths.

    Days 46–75

    Control testing

    Clinician accountability, subgroup performance and reconstruction tested against real episodes of care.

    Days 76–90

    Position and attestation

    Maturity position issued to the clinical governance committee, gap remediation sequenced, attestation scope agreed.

    Scope, residency and retention are set per engagement. Commercial terms are quoted against scope.

    Start a scoping conversation