Gauze builds intelligence for every part of the hospital: clinical care, operations, education and research. Models used in care are made to the standard of a medical device, and clinical knowledge is kept current with the evidence.

Work with us

A STEMI patient, from ambulance to discharge, seen four ways.

Illustrative case
01 / SERVE

Clinical Intelligence

The whole patient, in one view.

Following the patient to the next step.

    NoēsisEvidence layer
    1. Receive
    2. Review
    3. Return
    Evidence check

    Provenance, evidence and clinician oversight

    No check at this step

    Reviewed results return to the flow

    Clinical Intelligence: Following the patient to the next step.

    THE FRONTIER

    Medicine is an ever-changing science.

    New evidence revises the standard of care every year. Keeping a whole team current is hard, and practice drifts from the evidence in ways no one sees.

    Clinical AI has the same problem, more sharply. Most models never reach a patient; they stop at a paper or a pilot. Bringing one into care takes validation across sites, regulatory approval as a medical device, and monitoring for as long as it runs.

    Gauze treats models and clinical knowledge the same way. Both are built from evidence, tested before use and kept under review while in use. Models are made the way medicine expects any device to be made: to a standard, with evidence at every step, and watched after release. Clinical knowledge is held at its current standard and revised when the evidence changes.

    Four connected forms of hospital intelligenceClinical care, operations, education and research grids share a continuous evidence path. Signals follow grid lines and pass between the four areas, connecting patient care with institutional learning. Each platform contains a small display and a related record: vital signs and a patient record, capacity and a work queue, a case review with learning notes, and an evidence network and source library.Clinical careOperationsEducationResearch
    Connected dataEvidence in motion

    INTELLIGENCE

    Models improve across hospitals through federated learning. Patient data stays in the hospital where it was collected.

    The problem

    For a critical patient, vitals, imaging, model outputs and alerts sit in different systems. Assembling the picture takes time the patient may not have.

    Our approach

    Gauze brings these into a single view of the patient. Model outputs appear beside the source data they came from, and alerts show their escalation status. Guidance draws on the standard of care held in Noēsis, with its sources shown.

    The Intelligence Registry lists every model in clinical use, and whether each one is live, in shadow, paused or retraining. Noēsis monitors each one continuously and flags drift or weakening evidence.

    Fig. 01 / Clinical care

    PATIENT CONTEXTRECORDHistoryDiagnosesMedicationsAllergiesSIGNALSVitalsLab resultsImagingAlertsCARECare contextClinical notesCare plansHandoversPATIENT VIEWLive AI assessmentsNeurologicalLiveLive assessmentCardiacLiveLive assessmentECG viewerDetection overlayImaging viewerInference overlayAI findings alongside the sourceIllustrative modality viewAlert escalationFlaggedEscalatedAcknowledgedINTELLIGENCE REGISTRYLiveAssessmentsShadowSignalsPausedImagingRetrainingRisk modelNOĒSIS / CHECKS EVERY STAGEData qualityModel auditDrift checks

    EVIDENCE

    A model that works in a trial has not yet shown it works in a hospital. Gauze tracks every model and every piece of clinical knowledge as it earns, or loses, its place.

    Nothing stays applied by default. New evidence, outcomes or clinician feedback can send it back for revalidation. What no longer holds is withdrawn, and its record is kept.

    Noēsis reviews every move between stages. A model that is safe but adds cost or testing without helping patients moves back.

    Noemes in orbit around clinical truthModel noemes are circles and clinical knowledge noemes are diamonds. Both journeys run independently, at their own pace, through Conjectured, Tested, Validated, and Applied around clinical truth. Noēsis gates capture and ejection. Models can be demoted for drift or insufficient patient value, revalidated, or withdrawn with their failures on record. Knowledge can change practice at a preregistered threshold, or move outward when new evidence reverses a finding.CaptureEjectionConjecturedTestedValidatedAppliedClinicaltruth

    Closer to truth, never the final word.

    1. Conjectured. An idea with a question attached.

    2. Tested. Checked against real clinical data.

    3. Validated. Holds across settings.

    4. Applied. Cleared for use, or adopted into a clinical pathway.

    Noēsis boundaryEvidence orbitNoeme: modelNoeme: clinical knowledgeNewly capturedRevalidation / reversal / ejection

    Each unit in orbit is a noeme: a model or finding that carries its own provenance.

    FOUNDATIONS

    01 / Data layer

    Lattice

    Lattice connects every authorized source, including clinical systems, imaging, devices and operational records. It cleans, maps and aligns them into one consistent record over time. Every value can be traced to where it came from. Access is set by role and purpose, and data stays within the terms it was shared under.

    Lattice and Noēsis: two connected layersData travels along the upper Lattice grid. Continuous paths descend from grid intersections to the lower Noēsis layer, travel across its evidence grid, and rise back to Lattice. Both layers remain connected and visible.LatticeNoēsis
    Data through LatticeExchange with Noēsis
    02 / Evidence layer

    Noēsis

    Anything that reaches clinical use carries a record: its source, version, known limits and sign-off. Noēsis checks provenance, quality and approvals before anything advances. Once it is live, Noēsis monitors performance and flags drift or weakening grounds for use.

    Noēsis also holds the hospital's current clinical knowledge: the guidelines, evidence and protocols it practices by, each with its source and date. When the evidence changes, it flags what is affected, in care and in teaching.

    We work with hospitals, clinicians and researchers. Write to us if you want to take part.

    Work with us