Cardiac arrest & resuscitation
When the machine is wrong. Characteristics of true and false predictions of Out-of-Hospital Cardiac arrests in emergency calls using a machine-learning model
Study context
- Population
- Emergency calls to Copenhagen Emergency Medical Services (EMS), September 2018 to December 2019; verified Out-of-Hospital Cardiac Arrest (OHCAs) from the Danish Cardiac Arrest Registry
- Participants
- 169,236 emergency calls
- Outcome
- Characteristics of emergency calls with false-negative or false-positive OHCA predictions by the machine-learning model
Methodological detailssingle EMS (Copenhagen), False-negative and false-positive calls evaluated by trained auditors
- Single EMS (Copenhagen)
- False-negative and false-positive calls evaluated by trained auditors
Main finding
As reported in the abstract
Characteristics of false predictions
- Unrecognised OHCA: caller presented a condition completely different from OHCA
- 31%
- Unrecognised OHCA: patient reported breathing normally
- 28%
- Unrecognised OHCA: language barriers
- 23%
- False-positive OHCA: patient reported unconscious
- 28%
- False-positive OHCA: calls about dead patients with irreversible signs of death
- 13%
Further findings
| Outcome | Values |
|---|---|
| Model performance for OHCA recognition | Machine-learning model Calls suspected as OHCA 5,811 (3.4%) · Sensitivity 84.5% · Specificity 97.1% |
Original publication
When the machine is wrong. Characteristics of true and false predictions of Out-of-Hospital Cardiac arrests in emergency calls using a machine-learning model. Resuscitation. Published 9 January 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →