Paramedic PapersHome Cardiac arrest AI in cardiac arrest: oversight may be symbolic
Cardiac arrest
AI in cardiac arrest: oversight may be symbolic
Human-in-the-Loop in AI-assisted Resuscitation Protocols: Conceptual Pitfalls and Practical Considerations
Resuscitation Plus · 2026-11 · Added 8 November 2026
A conceptual analysis of AI-assisted OHCA resuscitation protocols argues that time pressure, protocol-driven practice and ambiguous authority can turn human-in-the-loop oversight of AI recommendations into symbolic governance rather than real crew control.
- Study type
- Other
- Design
- Conceptual/policy analysis (no empirical data)
- Participants
- Not reported
- Main finding
- No empirical data — proposes five policy measures including ecological trialling, designed-in explainability, defined escalation triggers and authority, and protection for justified deviation from AI recommendations
- Caveat
- Purely conceptual; no data on how crews actually behave with AI recommendations
- Bottom line
- Context only
Why it matters in Australia (Medium relevance). AI decision support is arriving in dispatch and clinical systems worldwide, and whether crews can meaningfully override it is a governance question every ambulance service will face.
Learning goal
Understand how authority, escalation triggers and justified deviation shape real oversight of AI-assisted resuscitation recommendations.
More
- Topic: Cardiac arrest
- Daily scan: 8 November 2026
- Podcast: Paramedic Papers Daily, 8 November 2026
Related papers on Cardiac arrest
-
PEEP during CPR linked with more ROSC
A single-centre Belgian cohort of intubated OHCA patients managed by a physician-staffed team found mechanical ventilation with PEEP during CPR was associated with higher odds of ROSC (adjusted OR 1.58) than manual ventilation without PEEP.
-
Mainstream and sidestream capnography give different ETCO2
A Spanish retrospective analysis of 23 OHCA recordings with concurrent mainstream and sidestream capnography found mainstream systematically recorded higher ETCO2 (bias 3.94 mmHg, limits of agreement -3.21 to 11.09 mmHg), though trends and abrupt rises/drops were consistently co-detected.
-
Community first responders trained but rarely dispatched
A cross-sectional survey of 2,704 registered community first responders in two Swedish regions before local system launch found most were CPR-trained and felt prepared (97.3%) but real dispatch experience was low (15.1% ever dispatched, 1.9% performed CPR).
-
CARE framework for traumatic cardiac arrest
A framework for integrating parallel priorities in traumatic cardiac arrest resuscitation — abstract not yet available, so no results to report.