Triaggent
Methodology

Grounded in ABA referral criteria.
Transparent by design.

Triaggent is built around the American Burn Association’s published referral criteria. It pairs those criteria with road-routed ground time and a modeled air-transport estimate, and surfaces a rationale alongside every ranked recommendation. This page summarizes the inputs, weights, and decision logic the current engine uses.

ABA-alignedRule-based scoringAuditable per case
01 — Criteria

Which ABA criteria trigger a burn-center transfer

The ABA lists the criteria below as indications for burn-center transfer. Triaggent’s current rule set scores a subset of these directly (TBSA bands, mechanism, inhalation, pediatric age); the remaining criteria appear in the intake narrative and rationale so the physician can weigh them alongside the ranked list.

  • Partial-thickness > 10% TBSA

    Any patient with partial-thickness burns greater than 10% total body surface area meets referral criteria.

  • Face, hands, feet, genitalia, perineum, major joints

    Burns involving these regions are referred regardless of size.

  • Full-thickness (3rd degree)

    Any third-degree burn in any age group is a transfer indication.

  • Electrical / lightning

    Electrical burns, including lightning injury, are referred independent of TBSA.

  • Chemical

    Chemical burns are referred; the receiving center is expected to have decontamination capability.

  • Inhalation injury

    Suspected or confirmed inhalation injury is referred, paired with Level 1 trauma capability where feasible.

  • Preexisting medical disorders

    Comorbidities that could complicate management, prolong recovery, or affect mortality escalate referral priority.

  • Burn + concomitant trauma

    If burn injury is the dominant risk, route direct to a burn center; if trauma dominates, stabilize at Level 1 trauma first.

  • Pediatric without peds capability

    Any burned child at a hospital lacking qualified pediatric personnel or equipment is referred.

  • Special social / rehab needs

    Cases requiring dedicated social, emotional, or rehabilitative intervention are referred to a verified center.

02 — Scoring

How each input is weighted

Scored intake fields map to a rule with an explicit weight. Weights are summed to produce the capability score that drives the initial ranking; transport time then breaks ties. No ML model decides the ranking — every number below is fixed and inspectable in packages/criteria-packs/burn/src/rules.ts.

TBSA thresholds
  • TBSA > 40%
    Top-tier weight; weights transport time heavily
    +10
  • TBSA 20–40%
    High-priority weight for ranking
    +8
  • TBSA 10–20%
    Meets ABA >10% partial-thickness threshold
    +6
  • TBSA < 10%
    Sub-threshold alone; other criteria may still apply
    +3
Mechanism + clinical flags
  • Chemical burn
    ABA referral criterion regardless of TBSA
    +8
  • Electrical burn
    ABA referral criterion; deep-tissue capability preferred
    +8
  • Inhalation mechanism
    Level 1 trauma capability preferred alongside burn center
    +7
  • Pediatric (< 18y)
    Pediatric-verified center preferred when available
    +7
  • Suspected inhalation injury
    Bronchoscopy and critical-care capability preferred
    +6

The composite rank is capability_score × 100 − transport_minutes. A higher-capability center can be outranked by a closer one only when the transport-time gap exceeds the capability gap, which keeps capability match dominant while still preferring faster routes when capability is equal.

03 — Transport time

How distance and travel time are determined

Every candidate center gets two transport estimates — ground and modeled air — computed in parallel, then the engine picks a mode.

Ground

OSRM road routing

Driving time and mileage come from OSRM (Open Source Routing Machine) over the live OpenStreetMap road network. We issue one route request per candidate center and use the returned duration and distance directly — no free-speed approximation.

  • Real road network, turn-by-turn feasible
  • Used as the baseline for every case
  • If OSRM is unreachable, the engine degrades to a straight-line fallback and marks the tier
Air

HEMS model

Air time is modeled, not dispatched. Great-circle distance (haversine) is divided by a 135 KTAS cruise speed, then a fixed 37-minute operational penalty is added:

air_min = (nm / 135 kts) × 60
        + 15 activation
        + 8 ground-to-launch
        + 6 landing
        + 8 handoff
  • Gated by HEMS base coverage of the incident
  • Gated by live NOAA METAR weather minimums
  • Never implies aircraft or crew availability
Mode selection
Air preference threshold20min
HEMS cruise speed135kts
Fixed air overhead37min

Air is only recommended when it saves at least 20 minutes over ground and passes both the coverage and weather gates. Otherwise ground is recommended. When both modes are feasible, the alternative is shown with its time delta so the coordinator can choose.

04 — Workflow

End-to-end decision pipeline

Every case follows the same pipeline. Inputs, rule hits, and the final ranked list are persisted so admins can review any recommendation after the fact.

  1. 1

    Intake → ABA criteria evaluation

    Age, TBSA, mechanism, inhalation, comorbidities, and location are validated. Rules fire against the ABA criteria; required capability tags (e.g. aba_verified_burn, pediatric_burn, level_1_trauma) are collected.

  2. 2

    Capability-matched candidate search

    Centers are queried with an expanding radius (100 → 200 → 400 → 800 mi) until a pool satisfying the required tags is found. If none match strictly, the engine relaxes the tag filter and marks a partial-match fallback tier.

  3. 3

    Dual-mode transport estimate

    For up to 5 candidates, ground (OSRM) and air (HEMS model) times are computed in parallel. Coverage and weather gates decide whether air is even on the table.

  4. 4

    Capacity signal cross-check

    The ranked list is enriched with cached capacity signals (beds available, source, last-updated). If a candidate has no capacity hit, or the known bed count is below the cohort size, it is marked 'call to confirm' so the coordinator can verify verbally.

  5. 5

    Composite rank + rationale

    Capability score (×100) minus transport minutes yields the final order. Each candidate carries a rationale line explaining which rules fired and which capabilities matched — surfaced to the coordinator and persisted with the case record.

Decision support, not a clinical substitute

Triaggent ranks and suggests; the coordinating clinician decides. Transport times are modeled estimates. The air model does not imply aircraft, crew, weather, or receiving-center availability. Capacity signals are cached and should be verbally confirmed before any transfer. This is a demo build; it is not a medical device and is not for clinical use.

See it applied to a case

Try the engine on a sample intake.

Enter age, TBSA, mechanism, and location. The rationale panel shows which rules fired and how transport time was selected.

Start triageBack to home