Skip to content
ESBook a demo
MenuClose
Book a demo

The ontology of health.

We read each person the way you read a tree: layer by layer.

In Colombia, a person's health information is split between their provider (IPS), the labs, the blood banks, their health plan (EPS) and their own phone. Arbol's ontology brings it together in seven layers, one on top of another, until the whole person is there: with the source of every data point and the permission to use it.

Request a meetingSee the seven layers
  1. 01SourcesWhere each data point lives
  2. 02Integration777+ connectors
  3. 03NormalizationOne language
  4. 04IdentityOne person, not four records
  5. 05OntologyObjects, relations and actions
  6. 06GovernanceWho sees what, and why
  7. 07The personTheir health DNA

Today, a person's health is in pieces.

Every provider, lab, blood bank and health plan keeps its own part, in its own format and sometimes under a different ID for the same person. Nobody is doing anything wrong: that's how the system is built. But the provider responsible for that person under a PGP contract (a prepaid, population-based contract) sees only a piece.

Example caseMarta, 29, Soacha. Week 24 of her pregnancy.

13health records4 of 13What her provider sees13 of 13What her provider sees with Arbol

  1. Picture Marta as the cross-section of a trunk: each ring, a stage of her life; each dot, a health record. She has thirteen.
  2. But those thirteen records live in seven places, and none of them sees what the others keep.
  3. Her provider is responsible for all of Marta under a PGP contract, and sees one piece: four of thirteen. The rest arrives late —as a PDF, in a portal— or never arrives.
  4. Two pieces never arrive: the HIV test from Cali and her blood donation. For her provider, Marta has no first-trimester screening: they chase what's already done, and the glucose test waits its turn.
  5. Arbol puts the pieces back together, each with its source and its permission; what has no connection she sends herself, as a WhatsApp photo nursing validates. Here's how, layer by layer:

Seven layers, one on top of another.

This is how we build an institution's ontology: each layer rests on the one below, and none erases what was underneath. We follow it with Marta's case. Example case

  1. 01 Sources

    Where each data point lives.

    We start with the inventory: the health record, lab, imaging, pharmacy, billing and RIPS, the health plan's authorizations, enrollment in the national registry (BDUA), and the phone and WhatsApp conversations. For each source we note its owner, format, frequency and who can authorize its use.

    Marta's sources
    • Health record · Clínica NorteAPIDaily
    • LaboratoryPDFPer result
    • PharmacyCSV fileWeekly
    • Health plan (EPS)PortalPer authorization
    • Calls and WhatsAppArbolReal time
    • A provider in CaliNot connectedShe brings it
    • Blood bankNot connectedHer donor slip

  2. 02 Integration

    We connect without moving anything.

    Each source comes in through its own connector —API, file, database or portal— with your institution's credentials. Your systems remain the source of truth: we read, we don't migrate. Every read records where it came from and when, and the same event is never processed twice.

    What came in on September 22
    1. 06:00Laboratoryfile3 results
    2. 06:02Health recordAPI1 visit
    3. 06:02Health recordAPIduplicate, dropped
    4. 07:14PharmacyCSV2 dispensings
    5. 19:03CallArbol1 conversation
    6. 19:06WhatsAppArbol1 photo

  3. 03 Normalization

    One language.

    Every source writes its own way: different dates, units, names and codes. We bring them to the standards: CIE-10 (ICD-10) for diagnoses, CUPS for procedures, CUM and ATC for drugs, LOINC for lab and DIVIPOLA for municipalities. A PDF or a photo of a result becomes fields, and whatever a machine can't be sure of is validated by someone on your team.

    Arrives like this · stays like this
    • Glicemia ayunas: 92 mg/dlFasting glucose · 92 mg/dLLOINC 1558-6
    • Hb 11,8Hemoglobin · 11.8 g/dLLOINC 718-7
    • SULFATO FERROSO 300MG TABFerrous sulfate · 300 mgATC B03AA07
    • SOACHA - CUNDINAMARCASoacha, CundinamarcaDIVIPOLA 25754
    • 03/06/262026-06-03
    • Photo: «VIH prueba rápida: no reactiva»HIV 1 and 2 · non-reactiveValidated by nursing

  4. 04 Identity

    One person, not four records.

    The same person arrives with a birth registration, a youth ID card, a national ID or a temporary protection permit, with a misspelled surname or a new phone number. We compare ID, names, date of birth, phone and municipality, and we only merge on evidence. When the evidence falls short, someone on your team decides, and every merge can be undone.

    Four records of Marta
    • RUIZ PÉREZ MARTA LUCÍACC ••• 4417 · 1997-03-14Clínica Norte
    • Martha RuízCC ••• 4417Laboratory
    • Marta RuizCell ••• 5521WhatsApp
    • Ruiz Perez Marta L.TI ••• 0932 · 1997-03-14Earlier provider, 2012
    • ID, date of birth and phone matchSame person
    • Different ID; names and date of birth matchMerged by nursing · Sep 12

  5. 05 Ontology

    Objects, relations and actions.

    Here data becomes a model. Everything in healthcare is an object with its properties and its source: the person, their enrollment, the contract that covers them, their care pathway, their visits, results, medications, appointments, conversations and consents. Relations say how they touch, and each object knows what can be done with it and who can do it.

    Person · Marta
    • enrolled inEPS · contributory plan
    • covered byPGP contract · Clínica Norte
    • on pathwayMaternal-perinatal · week 24
    • hasResult · HIV non-reactive · Cali
    • spoke viaCall · Sep 22, 7:03 p.m.
    • missingGlucose tolerance test · weeks 24–28

    What can be done

    • Book the glucose test
    • Confirm the checkup
    • Alert nursing

  6. 06 Governance

    Who sees what, for what, and with what permission.

    This is Ámbar, our governance. Every data point carries its purpose and its permission: the person's authorization is recorded per purpose, as Colombia's Law 1581 of 2012 requires; each role sees only what it needs, and every read lands in the audit log. A sensitive data point, like an HIV result, never travels in a reminder.

    Who sees her HIV result
    • Maternal program nursingSees the result
    • Her treating physicianSees the result
    • The scheduling assistantSees «screening on record»
    • BillingDoesn't see it
    • Arbol's teamDoesn't see it

    Her authorizations

    Care
    Authorized
    WhatsApp reminders
    Authorized
    Population studies
    Your institution decides

  7. 07 The person

    Their health DNA, in one view.

    On top sits the person: their timeline, pathways and cohorts, gaps and next step, every data point with its source. It isn't a report: it's what your team uses to decide whom to call today, and what Arbol uses to call, book and confirm.

    Her health DNA
    Marta
    29 · Soacha · pregnant, week 24
    Prenatal checkups
    4 of 4, up to date
    HIV screening
    First trimester: Cali, week 10 · validated
    Glucose tolerance test
    Due now · weeks 24–28
    Contact
    Prefers WhatsApp, after 6 p.m.

    6:10 p.m.Mónica, Clínica Norte's assistant, messaged her on WhatsApp. Glucose test booked: Thursday, 7:00 a.m., fasting.

  8. From above, the layers are rings.

    Seen from above, the stack is the cross-section of a trunk: the person at the center and every layer around her. Each new visit, result or conversation adds a ring. That is how a tree grows, and how what your institution knows about each person grows.

    • 01Sources7 sources
    • 02Integration5 connectors
    • 03Normalization6 coded values
    • 04Identity4 records, 1 person
    • 05Ontology6 relations
    • 06Governance3 purposes
    • 07The personMarta

What shapes her health.

Some data points aren't a result or a diagnosis, and they change everything: someone smoking at home, a soda every day, a mother with diabetes, a cousin with an orphan disease. Public health calls them determinants —the circumstances in which people are born, grow, live, work and age— and they are one of the pillars of Colombia's Ten-Year Public Health Plan 2022-2031. Arbol turns them into events, links them to the family and turns them into alerts for whoever cares for her. Example case

  1. 01 Ontology

    Every sentence becomes an event.

    «I have a soda with lunch, every day», Marta said on a call. «Lives with her father, who smokes», nursing wrote down. For Arbol, each sentence is a determinant event with its type, its person, its source and its date, and it stops getting lost in a note.

    Arrives like this · stays like this
    • «Me tomo una gaseosa con el almuerzo, todos los días»Habit · sugary drinks · dailyCall · Sep 22
    • «Los domingos salgo en bici de montaña»Risk activity · mountain bikingWhatsApp · Sep 14
    • «Convive con el papá, que fuma»Exposure · tobacco smoke at homeNursing note · wk 12
    • «Mi mamá es diabética»Family history · type 2 diabetes · motherPrenatal checkup · wk 8
    • «Un primo mío tiene hemofilia»Family history · hemophilia · maternal cousinCall · Sep 22

  2. 02 Topology

    One person's data touches others.

    Marta's health doesn't start with Marta. Her mother's diabetes is part of her history and her baby's; her father's smoking reaches the house where the baby will grow up; and a cousin's hemophilia, on her mother's side, says something about Marta that nobody has asked her. That map is the topology: who is whose child, who lives with whom, and where each event travels.

    Marta's connections
    • Rosamother ofMartaFirst-degree history
    • Jorgelives withMartaSame house, in Soacha
    • Juliáncousin ofMartaOn her mother's side
    • Martamother ofHer babyWeek 24

  3. 03 Knowledge

    We read it with what the country knows.

    Every event is checked against a clinical knowledge base our physicians maintain: the maternal-perinatal pathway in Resolution 3280 of 2018, the Ministry of Health's clinical practice guidelines, the official list of orphan diseases in Resolution 2625 of 2025, and the determinants in the Ten-Year Plan. That's how Arbol knows hemophilia runs through the mother's line, and that with a diabetic mother the glucose test can't wait.

    The rules that are met
    • Diabetes in the mother and pregnancy at week 24The glucose test, without delayRes. 3280 of 2018
    • Hemophilia, an orphan disease (Res. 2625 of 2025), in a cousin on the mother's sideMarta, a possible carrier; her baby, if a boy, could inherit itClinical genetics
    • Tobacco smoke at home and pregnancyCounseling for a smoke-free homeMaternal pathway
    • Mountain biking and pregnancyCounseling on safe physical activityMaternal pathway

  4. 04 Alerts

    A system that warns ahead.

    We don't wait for Marta to show up with a problem. Every rule that is met becomes an alert for the right person on your team, with the why, the source of each data point and a deadline. What gets resolved closes itself; what goes overdue moves up.

    Marta's alerts, today
    1. 1Book the glucose tolerance testMother with type 2 diabetes · soda every dayMaternal program nursingBy Oct 20
    2. 2Assess genetic counseling and carrier testingA cousin on her mother's side with hemophiliaHer obstetricianNext checkup
    3. 3Counseling: a smoke-free homeLives with a smokerNursingNext checkup
    4. 4Safe physical activity in pregnancyMountain biking on SundaysHer physicianNext checkup
    5. 5Family history, ready for pediatricsDiabetes and hemophilia in the familyPediatricsAt birth

  5. 05 Governance

    Every connection, with its permission.

    What Marta tells us about her family is Marta's data, and it's used in her care because she authorized it. Her mother's health record belongs to her mother: it's linked only if Rosa authorizes it. And no alert reaches Marta as a diagnosis or a risk: it reaches her care team, and the professional decides.

    What is used, and with what permission
    • What Marta told us about her familyUsed in her care
    • Rosa's health recordOnly if Rosa authorizes it
    • Julián's health recordNot used

    Authorizations

    Her care
    Authorized by Marta
    Linking Rosa's record
    Pending Rosa
    Alerts
    Only for her care team

You decide what we study.

Some institutions ask us to operate and nothing more; others authorize us to study their population. Both are valid: the difference is put in writing, and it looks like this.

Who sees what, depending on what you authorize
Who sees what, depending on what you authorize
IdentityClinical dataConversationsIndicators
Your clinical teamSeesSeesSeesSees
Your administrative teamSeesWhat's neededSeesSees
Arbol's agentsThe minimumNoTheir ownNo
Arbol's teamNoPseudonymNoPseudonymNoPseudonymNoSees
Third-party AI modelsNever train on your data. Zero retention.
  • We use the data only for what was contracted: answering, booking, confirming and reporting.
  • No one at Arbol studies your population. Support happens on the platform, with an audit trail, and no one downloads your patients' data.
  • Our physicians, nurses and analysts study your population under pseudonyms, inside your institution's environment.
  • What we find —cohorts, gaps, risks— is yours and delivered in writing. The authorization has a scope and a date, and it can be withdrawn.

Either way, your data lives in a database dedicated to your institution, is never sold and never trains third-party models.

A common model for health in Colombia.

Every institution on Arbol shares the same ontology: the same objects, the same relations, the same codes. What they never share is data: each one has its own database. That's why a gap or an indicator means the same thing in a provider in Soacha and in a network in Barranquilla.

Colombia has taken the first step: every provider sends the national platform a Digital Care Summary of each consultation, emergency visit or hospital stay. Our ontology speaks that language and adds what a care summary doesn't carry: the call, the message, the donation, the family, the contract and the permission.

The country's language

Resolution 1888 of 2025
The Digital Care Summary (RDA), on HL7 FHIR R4, for every consultation, emergency visit or hospital stay.
Law 2015 of 2020
The interoperable electronic health record.
Resolution 2275 of 2023
RIPS in JSON, as support for the electronic invoice.
Law 1581 of 2012
Personal data and the data subject's authorization.

  • PersonDIVIPOLAID, date of birth and municipality

    • EnrollmentBDUAHealth plan, regime and status

      • ContractDecree 441 of 2022PGP, capitation or fee-for-service, with its technical note

    • FamilyKinship and household, with each person's permission

    • Determinant eventResolution 1035 of 2022Habit, exposure, history or condition, with its source

    • Pathway and cohortResolution 3280 of 2018Maternal-perinatal, chronic care, prevention

      • AlertA gap, a critical result or a determinant event, with an owner and a deadline

    • VisitHL7 FHIR R4Consultation, emergency or hospital stay

      • DiagnosisCIE-10

      • ProcedureCUPS

      • ResultLOINC

      • MedicationCUM

    • ConversationCall, WhatsApp, text, email or web, with its reason and outcome

      • AppointmentSite, professional and status

    • ConsentLaw 1581 of 2012Purpose, date and channel

What we never do with a data point.

  • Sell it, or share it with advertisers.
  • Train third-party models on your patients' data.
  • Study your population without your written authorization.
  • Merge two people without evidence: a doubt is settled by someone on your team.
  • Link a relative's health record without their authorization.
  • Overwrite it: every value keeps its source and its history.
  • Tell a patient a diagnosis or a risk. Alerts are for your team, and the professional always decides.

Let's start with your data.

A meeting with our physicians, nurses and data engineers to go through your sources, your contracts and what your institution could see of each person.

Request a meetingArbol EnterpriseTrust Centersupport@getarbol.com

In Bogotá, at any hour. Arbol is answering.

Phone, WhatsApp, text, email and web chat.

Let's look at your data, layer by layer.

Book 30 minutes with our team.

A meeting with our physicians, nurses and data engineers about your sources, your contracts and what your institution could see of each person.

Book 30 minutesMessage us on WhatsApp
Or pick a day

support@getarbol.com

In Bogotá, at any hour. Arbol is answering.