DECISION SUPPORT ONLY · NOT A DIAGNOSTIC SYSTEM · CLINICAL JUDGMENT REMAINS WITH THE HEALTH WORKER
Attica region pilot · Greece

Clinical guidance
where there is no clinical guidance.

TriageOS is an offline-first clinical decision support system that runs on a $40 smartphone in a zero-connectivity zone. It guides community health workers through WHO IMCI protocols — what to ask, what to record, when to refer.

Draft v0.1 — Unvalidated. For research and testing purposes only.
0
Total triages
0
Today
0
Referrals
0
NGO partners
WHY THIS EXISTS
10M
health worker shortage by 2030 (WHO)
2.6B
people globally with no internet access
70%
of frontline health interventions delivered by CHWs

The world has digitized hospital care. It has not digitized the village health post. TriageOS changes that.

HOW IT WORKS
From village to protocol in three steps
01
CHW opens a link
No app store. Open a URL in any browser. TriageOS installs itself as a PWA and works completely offline from that point forward.
02
Guided clinical interview
WHO IMCI protocols run as an adaptive questionnaire. The system guides what to ask next based on each answer.
03
Decision + encrypted sync
Clear output: treat, refer, or urgent. Data encrypted locally. Syncs to the dashboard when connectivity is available.
Offline-first
Full functionality with no network. Caches everything locally. Syncs automatically when a signal appears.
Bilingual
Full English and Greek interface. Clinical terminology accurate in both languages.
WHO IMCI-referenced
Decision support only — not diagnostics. Protocol logic references established WHO IMCI triage methodology. Pending clinical review.
Transparent reasoning
Every recommendation shows exactly why. Full audit trail. No black box decisions in healthcare.
Outbreak detection
Rule-based anomaly detection across triage records. Early geographic clustering alerts before hospital admissions rise.
OpenMRS compatible
FHIR R4 integration layer built for OpenMRS. Triage data maps directly into existing NGO clinical records.
PLATFORM COMPARISON
The gap existing platforms leave open
Feature
OpenMRS
CommCare
Medic
TriageOS
Works with zero internet
Instant PWA / URL deploy
Pre-built WHO IMCI logic
Partial
On-device AES-256 encryption
Auditable decision logic
Partial
OpenMRS FHIR integration

The clinical gap in the world's most disconnected places is a software problem.

TriageOS is open source. The mobile app is free for volunteer health workers. Enterprise tiers for health ministries and international NGOs with sync dashboards and OpenMRS integration.

Live overview
Loading real data...
LIVE
Total triages
Referrals
Today
Referral rate
Outcomes breakdown
Intelligence alerts
Loading...
Recent triages
CHWProtocolAge groupOutcomeTime
No triages yet
Childhood Pneumonia
Protocol P-01 · Built on WHO IMCI guidelines · Age 2mo – 5yr
Childhood pneumonia
Offline
P-01 · WHO IMCI Step 1 of 5
Age
RR/min
Danger
Step 1 — Age
How old is the child?
Age determines the fast-breathing threshold. This affects the entire protocol — count carefully.
Step 2 — Main symptom
Does the child have cough or difficulty breathing?
Entry criterion for pneumonia assessment. If absent, pneumonia is unlikely — assess for other conditions.
Step 3 — Breathing rate
Count breaths for exactly 60 seconds
Fast breathing threshold for this age: 50+ breaths/min
60
Tap start, then count each breath
Step 4 — Danger signs
Does the child have ANY of these?
Check each carefully. ONE danger sign = refer immediately, regardless of breathing rate.
Step 5 — Record
Document this encounter
Record
0 records queued
Malaria Triage
Protocol P-02 · WHO IMCI · Origin-aware routing
Malaria assessment
Offline
Before you begin — P-02
Patient's country of origin
TriageOS automatically selects the correct WHO protocol branch based on the patient's origin country endemicity profile.
Select country / region
P-02 · Malaria · WHO IMCI
Protocol library
Referencing WHO IMCI triage methodology · Draft — pending clinical review
Childhood pneumonia
WHO IMCIv0.1 draft
Malaria triage
WHO IMCIv0.1 draft
Maternal danger signs
WHO ANCComing soon
Sepsis red flags
Sepsis-3Coming soon
Malnutrition screening
WHO CMAMComing soon
Neonatal danger signs
WHO IMNCIComing soon
Outbreak detection
Rule-based geographic anomaly detection · Grows with field data volume
Active insights
Awaiting field dataPENDING
Outbreak alerts will appear once triage records are collected. Detection uses deterministic rule-based logic — not probabilistic AI — to flag geographic symptom clusters.
Training module — in developmentROADMAP
Mini-lessons and simulations for CHWs who repeatedly struggle with specific conditions. Levels up workers over time.
Human escalation — in developmentROADMAP
One tap connects the CHW to a supervising clinician for dangerous or unusual cases. All final clinical decisions remain with the qualified health worker.
Data coverage
Childhood pneumonia
Alert sensitivity improves with triage volume. More records enable finer geographic cluster detection.
System limitations
Input quality: AI is only as good as the information entered. Missing or incorrectly described symptoms may produce unreliable recommendations.
Language and culture: Symptoms are not described the same way everywhere. Local context matters and cannot always be captured in a protocol.
No hardware integration yet: Symptom descriptions alone are less reliable than measured vitals. Future versions will support thermometers and pulse oximeters.
Legal responsibility: TriageOS provides decision support only. The CHW, supervising clinician, and partner organization retain full clinical responsibility.
About TriageOS
Open-source clinical decision support for community health workers
What this is
TriageOS is a decision-support tool — not a diagnostic system. It provides flowchart-style guidance referencing WHO IMCI triage methodology to help community health workers know when to refer, what questions to ask, and what vitals to record. The clinical judgment always remains with the health worker.
What this is not
TriageOS is not a diagnostic tool, a medical device, or a replacement for clinical training. It does not produce diagnoses. It does not replace physicians, nurses, or medical supervision. Following its output does not guarantee correct management.
Deployment
RegionAttica, Greece
ProtocolP-01 v0.1 draft · P-02 v0.1 draft
Clinical reviewerPending
StatusDraft — not validated
Built byAmrin Majumder
School21st Gymnasio of Athens
Privacy Architecture
Clinical data is encrypted on-device using AES-256-GCM before storage. Every record is chained with a SHA-256 hash linking it to the previous entry — creating a tamper-evident local ledger. No raw patient data is transmitted to any server.
EncryptionAES-256-GCM (WebCrypto)
Key derivationPBKDF2 · SHA-256 · 100k iterations
Audit trailSHA-256 hash chain
PHI to serverNONE
GDPR scopePseudonymised tokens only
Patient identityHMAC-SHA-256 token · device salt
Roadmap
+Voice input and offline audio guidance for low-literacy contexts
+Hardware integration — thermometer, pulse oximeter, BP monitor
+Human escalation — one tap to nurse or doctor hotline
+Local disease adaptation by region, season, and geography
+Built-in CHW training modules and competency tracking
+Arabic, Farsi, French, Tigrinya language support
+Contactless vitals — camera-based heart rate estimation (rPPG)
+Offline device-to-device record sync via QR bundle export
Clinical Assessment
AVPU + MUAC + CRT + Breathing timer · Zero equipment · Fully offline
Clinical Assessment
Offline
AVPU CONSCIOUSNESS SCALE
CAPILLARY REFILL TIME (CRT)
Press fingertip for 5 seconds, release. Count seconds until colour returns.
MUAC — MID-UPPER ARM CIRCUMFERENCE
Measure left arm midpoint. For children 6 months to 5 years.
60-SECOND BREATHING COUNTER
60
Count each breath. Enter count in resp rate above.
TEMP C (optional)
RESP /min (optional)
Outbreak Intelligence
Live triage pattern analysis · Cluster detection · Offline capable
Outbreak Intelligence
Analysing
TOTAL
REFERRALS
REF RATE
PROTOCOL DISTRIBUTION
OUTCOME FREQUENCY
Run triages to activate outbreak intelligence
Install TriageOS
Works offline · No app store