Home Operations & systems AI telephone triage protocol for EMS calls
Operations & systems
AI telephone triage protocol for EMS calls
trIAje project: protocol for a retrospective cohort study to optimise AI-assisted telephone triage of time-sensitive conditions in emergency medical services
BMJ Open · 2026-09-28 · Added 6 October 2026
Protocol for the trIAje project, a Spanish retrospective cohort linking 2020–2024 dispatch records, ePCRs and health-registry data to evaluate and then improve AI-assisted telephone triage of time-sensitive conditions using NLP on dispatcher call text.
- Study type
- Other
- Design
- Retrospective cohort study protocol
- Participants
- All calls 2020–2024 coded as unconsciousness, cardiac arrest, respiratory difficulty, chest pain or suspected stroke (number not stated)
- Main finding
- No results yet — will compare machine-learning triage performance against final clinical severity, reporting sensitivity, specificity, under- and over-triage and sex-stratified disparity checks
- Caveat
- Protocol only — no performance data; retrospective Spanish dispatch data may not generalise
- Bottom line
- Context only
Why it matters in Australia (Medium relevance). A trial protocol rather than a result — telephone triage under- and over-triage is the same problem Australian dispatch services face, so this signals what's coming.
Learning goal
Understand how AI-assisted telephone triage may affect under-triage of time-sensitive conditions and sex-based disparities in dispatch care.
More
- Topic: Operations & systems
- Daily scan: 6 October 2026
Related papers on Operations & systems
-
On-site CT slashes event transports
Retrospective before–after study at the Munich Oktoberfest found an on-site mobile CT led to 429 of 498 imaged patients (86%) being treated and discharged on-site, avoiding an estimated 22,737 minutes of EMS deployment time.
-
EMS dashboards: thin crisis evidence
Scoping review of 26 studies found EMS performance dashboards are used almost exclusively for routine operations — only one study addressed crisis conditions and it used no dashboard.
-
Patients report prehospital safety incidents
Structured telephone interviews 15–20 days after EMS transport across eight Spanish provincial services characterised patient safety incidents reported by patients and witnesses, covering near misses and no-harm events as well as harm.
-
What paramedics want from voice-to-text ePCRs
A mixed-methods survey of 323 Canadian paramedics plus 14 interviews identified and ranked the features paramedics want in speech-recognition documentation — 86% viewed it positively, with accuracy, patient interaction and compliance rated most important.