---
title: >-
  Copenhagen Emergency Medical Services: Real-time ML listening and dispatcher
  cardiac-arrest alerts
slug: copenhagen-ohca-dispatch-rct
stable_id: bfb4dcb5dde322ef
company: Copenhagen Emergency Medical Services
function_code: engineering
pattern_codes:
  - creator_to_judge
evidence_strength: verified
publication_tier: showcase
freshness: current
reviewed_at: '2026-08-23'
updated_at: '2026-08-22'
source_quality_summary: 1 peer reviewed; publication outcomes are verified.
caveat_summary: >-
  High-quality randomized negative result. The model alone was more sensitive but
  materially less specific and had lower positive predictive value than dispatchers.
collections:
  - negative-results
bundle_version: 1.0.0
bundle_fingerprint: sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93
canonical_url: https://brianletort.ai/transformations/copenhagen-ohca-dispatch-rct
---

# Copenhagen Emergency Medical Services: Real-time ML listening and dispatcher cardiac-arrest alerts

A live emergency-call alert may not improve dispatcher recognition despite higher standalone model sensitivity.

Function: Engineering design. Patterns: Creator to judge. Evidence: verified.

Freshness: current. Reviewed: 2026-08-23. Updated: 2026-08-22.


Source quality: 1 peer reviewed; publication outcomes are verified.

## Before

1. **Medical dispatcher** — Listen to the caller, recognize suspected cardiac arrest, and initiate the protocol. (control: Standard dispatch protocol)
2. **Dispatcher** — Coach CPR and dispatch resources when cardiac arrest is recognized. (control: Human clinical judgment)

## After

1. **Speech-recognition ML model** — Listens to ongoing calls and flags suspected cardiac arrest. (control: Model alert)
2. **Medical dispatcher** — Considers the alert and decides whether to recognize OHCA and initiate the standard response. (control: Dispatcher retains all response rights)

## Decision rights

The model alerts; the dispatcher decides whether to declare suspected OHCA and start CPR coaching and dispatch.

## Exception path

Dispatchers ignore false alerts and follow standard protocols; no response is automatically triggered by the model.

## Outcomes

- **Dispatcher recognition of confirmed out-of-hospital cardiac arrest** (verified): 90.5% with standard protocol and no model alert → 93.1% with the alert, P=.15; no statistically significant improvement. High-quality randomized negative result. The model alone was more sensitive but materially less specific and had lower positive predictive value than dispatchers.

## Executive lesson

A better classifier is not a better operating model. The alert's low precision and insertion into a time-critical conversation prevented technical sensitivity from becoming human performance.

## Anti-pattern

Selecting on model sensitivity while ignoring alert precision and operator response.

## Questions for leaders

- Does the alert improve the human decision?
- What false-positive rate is tolerable in a time-critical queue?

## Sources

- [Effect of Machine Learning on Dispatcher Recognition of Out-of-Hospital Cardiac Arrest During Calls to Emergency Medical Services](https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2774644) — JAMA Network Open