Acute cardiac care & resuscitation
Machine learning to identify hypoxic-ischemic brain injury on early head CT after pediatric cardiac arrest
Authors' conclusion
Deep learning models can identify HIBI on early CT images after pediatric OHCA and detect some presumed HIBI visually not identified by a radiologist.
Study context
- Population
- Children with CT within 24 h of out-of-hospital cardiac arrestOut-of-hospital cardiac arrest (OHCA) cases analysed
- Participants
- 117 children
- Comparison
- OHCA cases vs Controls
- OHCA casesn = 117
- Outcome
- Model discrimination of HIBI (OHCA cases vs controls) on early CT, by AUC
Methodological detailsage-matched controls, deep learning
- Age-matched controls
- Deep learning
Performance
As reported in the abstract
Mean AUC, OHCA cases vs controls
0.87 ± 0.05
Further findings
Model-identified presumed HIBI among 73 OHCA cases without radiologist-identified HIBI · death · values are death rates
P = 0.174
- Model-identified presumed hypoxic-ischemic brain injury (HIBI) among 73 OHCA cases without radiologist-identified HIBI — Model identified presumed HIBI
- 36 %
Prediction of death and unfavourable outcome (PCPC 4-6 at discharge) among OHCA cases · hospital discharge
- Mean AUC, death
- 0.79 ± 0.06
- Mean AUC, unfavourable outcome
- 0.69 ± 0.06
- 43 % died; 58 % unfavourable outcome
Discrimination of OHCA cases with radiologist-identified HIBI (n=44) from controls
- Mean AUC
- 0.98 ± 0.01
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Summarised from the published abstract; every number is checked against it