Pediatric anesthesia & critical care
Machine learning to predict poor school performance in paediatric survivors of intensive care: a population-based cohort study
Authors' conclusion
A machine learning model using data available at time of ICU discharge predicted failure to meet minimum educational requirements at school age.
Study context
- Population
- Children < 16 years surviving intensive care unit (ICU) admission in Queensland, Australia, 1997–201937,200 National Assessment Program-Literacy and Numeracy (NAPLAN) tests
- Participants
- 13,957 children
- Outcome
- Failure to meet the National Minimum Standard (NMS) in NAPLAN assessment during primary and secondary school
Methodological detailsprediction model development, machine learning, stratified nested cross-validation, Median follow-up 6 years
- Prediction model development
- Machine learning
- Stratified nested cross-validation
- Median follow-up 6 years
Main finding
As reported in the abstract
Failure to meet NMS in NAPLAN (prediction)
| Group | AUROC | Specificity at 85% sensitivity | Relative Area Under the Precision Recall Curve (rel-AUPRC) |
|---|---|---|---|
| Model | 0.8 (SD 0.01) | 51% | 3.42 (SD 0.06) |
Model
AUROC0.8 (SD 0.01)
Specificity at 85% sensitivity51%
Relative Area Under the Precision Recall Curve (rel-AUPRC)3.42 (SD 0.06)
Observational data: associations, not effects.
Further findings
| Outcome | Values |
|---|---|
| Model performance in children with no comorbidities admitted 2009–2019 (subgroup) | model AUROC 0.77 (SD 0.03) · rel-AUPRC 3.31 (SD 0.42) |
Failure to meet National Minimum Standard (NMS) by school grade (3, 5, 7, 9)
- Grade 3
- 14.7%
- Grade 5
- 17%
- Grade 7
- 15.6%
- Grade 9
- 16.6%
Main contributors to predictions
- Socio-economic status, illness severity, and neurological, congenital and genetic disorders contributed most
Original publication
Machine learning to predict poor school performance in paediatric survivors of intensive care: a population-based cohort study. Intensive Care Med. Published 24 June 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →