Aequus

Clinical AI

Aequus · Latin · equal

Equitable health is an information problem.

Before it is a technology problem. There is no shortage of health data. What is missing is context, and without it data never becomes something anyone can trust, understand or act on.

HaloPanel view
  • A1c 8.2%Drawn 14 months ago · outside lab
  • Statin not filled3 refills missed · pharmacy claim
  • No transport on fileSDOH · community screening
  • Interpreter needed, ArabicRegistration · 2024
  • Panel median A1c 6.9%412 patients · this clinic

Every line carries where it came from. Read only, over FHIR.

Our approach

Health systems are not short of data. Every institution we work with is drowning in it: screenings, claims, notes, registries, wearables, referrals. What none of it carries is context, so it arrives hard to find, too complex, untimely and untrusted, and the gap widens fastest for the people already furthest from care.

The industry's answer has been to add another system. Another login, another dashboard, another copy of the record sitting outside the institution's walls. That is why adoption stalls: it asks a clinician to leave the workflow they are in, and asks a compliance officer to accept data leaving the building.

We build the context layer instead. Signal is pulled from what already exists, given provenance, population, place, point of care and purpose, then surfaced inside the workflow the person is already in. It is read only over HL7 FHIR, it runs on the institution's own infrastructure, and the tuned model is theirs. Nothing duplicated, nothing leaves.

The Halo Suite

CompanionPlain-language answers grounded in the person's own record and their clinic's protocols. Not a chatbot bolted on beside care.For patients
PanelThe clinician sees the population behind the patient in front of them, inside the workflow they are already in.For the panel
CommunityCulturally-fit guidance and SDOH follow-through, closing the loop from clinic to community. Built with FQHCs and local clinics in mind.For the community
ResearchCohort discovery and real-world evidence without data leaving the institution's walls. Audit trails and standards throughout.For research

It augments the EMR. It does not replace it. Read only: nothing duplicated, nothing leaves.

How we build

01Pull

Signal from what already exists. Screenings, notes, claims, community data. Nothing new to collect.

02Contextualize

Attach provenance, population, place, point of care and purpose. This is the step everyone skips.

03Surface

Deliver it inside the workflow the person is already in, not beside it. Not one more system to log into.

04Return

Capture what the human did with it and feed that back in. The human stays in the loop at every turn.

A closed loop inside the human workflow. The loop returns to step one.

Adoption on your terms

Phase 0An open-weight clinical model on your cloud, grounded by a versioned knowledge base of your own protocols. Your cloud, your jurisdiction.Grounded
Phase 1A supervised fine-tune on your de-identified corpus. The tuned model is your artifact, on your endpoint.Tuned
Phase XThe same tuned model on an appliance inside the hospital. Your building, offline capable.On premise

Start where you are comfortable. Stop where the value is.

Built on standards

  • HL7 FHIR R4
  • SMART on FHIR
  • SNOMED CT
  • GS1 Digital Link
  • ISO 27269

AqTrueSource resolves one barcode to the right answer for whoever is holding it. A recalled batch resolves to a superseded notice, a flagged serial to an unverified-product page, the same GTIN in the Gulf to the Arabic leaflet.

krishi@aequus.health