CLARYS Intelligence

From your documents to your patients’ medical history.

With specialized semantic intelligence, CLARYS Intelligence identifies, structures and contextualizes the medical information scattered across your documents.

A multilingual, resource-efficient and traceable technology, built on reference medical and social terminologies. Available in CLARYS Medical Memory or directly in your applications.

Evaluate CLARYS Intelligence
type 2 diabetes mellitus
Label
type 2 diabetes mellitus
System
SNOMED CT
Code
44054006
Hierarchy
Clinical finding

1 mention

#1 · page 1 · type 2 diabetes mellitus

0.377 · 0.101 · 0.153 · 0.013

Patient referencesCopy JSON
What sets us apart

Understand clinical meaning, not just recognize words.

Non-generative, explainable, frugal AI that understands medical language.

Specialized clinical intelligence

Concepts, negation, uncertainty and medical history: clinical meaning, not just words.

One expression, several meanings
“MS” in the cardiology note, read as mitral stenosisDisambiguated
Type 2 diabetes, known since 2018Confirmed
No diabetic retinopathyRuled out
Suspected early nephropathySuspected
History of cholecystectomy in 2010Past history
Father has type 2 diabetesFamily history

Reference terminologies

Concepts aligned with SNOMED CT and ICD, in every language: a shared, verifiable clinical interpretation.

One concept, several languages
FRdiabète de type 2
ENtype 2 diabetes mellitus
DEDiabetes mellitus Typ 2

Reference concept

SNOMED CT 44054006ICD-11 5A11

Verifiable traceability

Every piece of information links back to the passage that supports it.

metforminSNOMED CT · 372567009
Source mention
Consult notePage 1 · line 3

Current treatment: metformin 1000 mg twice daily.

Open at mention

Medical and social understanding

Medical and social information combined to make sense of complex situations.

Social determinants of health
Illustrative example
Living conditionsLives alone, 3 floors up, no elevator
Social supportPrimary caregiver: her daughter
OccupationRetired, former textile worker

Clinical extraction

Identify the medical concepts in supported documents.

Consult note

56-year-old patient under follow-up for type 2 diabetes mellitus and hypertension.

Workup in March 2024 revealed iron deficiency anemia.

Current treatment: metformin 1000 mg twice daily.

The patient reports atypical chest pain on exertion.

Conclusion: continue follow-up, recheck glycated hemoglobin in three months.

SNOMED CT · 44054006Page 1

7 concepts · 7 mentions

Clinical finding Procedure Pharmaceutical or biologic product Observable entity

Explainable and reproducible

Deterministic extraction, with no generative model: same document, same result.

type 2 diabetes mellitus
Label
type 2 diabetes mellitus
System
SNOMED CT
Code
44054006
Hierarchy
Clinical finding

1 mention

#1 · page 1 · type 2 diabetes mellitus

0.377 · 0.101 · 0.153 · 0.013

Patient referencesCopy JSON
In production

Real time. Large volumes. Always up to date.

A specialized engine built for everyday clinical use, from a single consultation to entire archives.

  • Real time. 120 ms response time: clinical information at hand during the consultation, as you navigate.

  • Large volumes. 10,000 patients ingested in under 2 minutes (excluding OCR): entire archives, usable right away.

  • Incremental indexing. Each new document is indexed on arrival, without reprocessing existing content: knowledge stays up to date.

  • Resource efficiency. Runs on CPU, no GPU required: a deployment sized to your needs.

Incremental indexing
Continuous
  • ED note2 s agoIndexed
  • Lab results1 min agoIndexed
  • Office visit note4 min agoIndexed

New documents processed on arrival, without reprocessing existing content.

For care

The medical information that matters, surfaced from documents.

Identify documented diagnoses, treatments and history, with their clinical context and direct access to the source passages.

  • Annotated viewer. Clinical concepts highlighted in the document, for faster reading.

  • Search the content. Medical information, beyond titles and dates.

  • Coordinate a complex care pathway. Combine clinical information with documented social and functional data.

Consult note

Patient under follow-up for type 2 diabetes and hypertension. Treatment mentioned: metformin.

Information identified

  • Type 2 diabetesConfirmed
  • HypertensionConfirmed
  • MetforminTreatment mentioned

Source: consult note, page 1

Clinical Case Search

Similar patients. Sometimes different pathways.

Search for patients by clinical characteristics and explore similar care pathways to lay the groundwork for research and population analysis.

  • Search by clinical criteria. Build cohorts from clinical and temporal criteria and, where available and authorized, social criteria.

  • Comparable trajectories. Compare trajectories by events, treatments and conditions, to spot similar pathways and understand what sets them apart.

Trajectories and populations
Clinical Case Search
Illustrative example
HER2+ breast cancer
  • HER2-positive breast cancerClinical finding
  • Hormone receptor positiveObservation
  • TrastuzumabProduct
#1
#2
#3
#4
30 y45 y60 y75 y

Mentions by patient age

214 records match the searchAge at first mention: 32 to 78 · 4 facilities
Request an extraction
For research and data

More clinical information. A more complete view. More relevant analyses.

Much medical knowledge lies dormant in documents. CLARYS Intelligence makes it usable, structured and traceable, without adding new sources.

  • Study feasibility. Find eligible patients in clinical reports.

  • Enriched cohorts. Inclusion criteria enriched by free text.

  • Real-world data. Better-characterized comorbidities, treatments and events.

  • Longitudinal analyses. Event sequences ready for care pathway studies.

From source to variable
VariableContextCodeSource
Type 2 diabetesConfirmedSCT 44054006Consult note · p.1
Diabetic nephropathySuspectedSCT 127013003Consult note · p.2
Diabetic retinopathyRuled outSCT 4855003Ophthalmology report · p.1

Each variable remains linked to its original passage.

For management

Clinical intelligence built for production constraints.

Quality, documented

Precision, recall and coverage measured on your corpus.

Fits your existing stack

CLARYS Intelligence sits between your content and your warehouse, without replacing your pipelines.

FHIR R4

Standardized resources, depending on available profiles.

For software vendors and integrators

Enrich your applications. Without reinventing clinical intelligence.

You know your users. We bring clinical intelligence, aligned with recognized reference terminologies, without tying up your teams building a semantic engine.

  • Enrich your features. Extraction, context and semantic search, built in.

  • Reduce development time. Your teams focus on what differentiates your product.

  • Open up new uses. Patient context, richer data, third-party services.

Vendor integration
API response
200
{
  "lang": "en",
  "mentions": [{
    "surface": "metformin",
    "codes": ["SCT:372567009", "ICD11:XM0JN5"],
    "fragment_span": { "start": 214, "end": 223 },
    "doc_anchor": { "left": 0.21, "top": 0.31 }
  }]
}
Concrete applications

One clinical intelligence. Many applications.

Unlock your document archives

Make your archives speak: clinical information, finally searchable.

Archives and Intelligence

Enrich a data platform

New variables, drawn from your clinical reports.

Clinical data warehouses and semantic interoperability

Understand populations

Cohorts and longitudinal analyses built on routine care data.

Trajectories and populations

Support complex care pathways

Clinical and social context that follows the person across organizations.

Care pathway continuity

Key takeaways

Medical language, understood and structured, with a link to every source.

  1. Meaning, not just words

    Negation, uncertainty, family history: every mention read for what it is.

  2. Reliable and explainable

    Deterministic extraction, no generative model, aligned with SNOMED CT and ICD.

  3. Fast and frugal

    Real time and large volumes, on CPU, no GPU required.

Standalone use

Your environment. Our clinical intelligence.

Archive, CDW, line-of-business software or data platform: let’s test CLARYS Intelligence on your content and key variables.

Test a corpusScope an integration