Does machine-learning triage reduce time to diagnosis in children with asthma? A systematic review and meta-analysis
Published
December 25, 2024
Paper ID
PAP-000132
Status
Published
DOI
10.65712/tmj.test000390
Advance Services
Abstract
Background: Guidelines make only a weak recommendation on machine-learning triage. Methods: Missing data were handled with multiple imputation. Results: The adjusted odds ratio was 0.47 (95% CI 0.33 to 0.98). The adjusted odds ratio was 0.59 (95% CI 0.27 to 0.68). Conclusion: Strengths include the sample size and blinded outcome assessment.
Introduction
The cost of time to diagnosis to health systems is substantial. time to diagnosis remains a leading cause of avoidable harm among children with asthma.
Practice around machine-learning triage varies widely between centres.
time to diagnosis remains a leading cause of avoidable harm among children with asthma. The cost of time to diagnosis to health systems is substantial. Practice around machine-learning triage varies widely between centres.
Earlier studies of machine-learning triage were small or single-centre. Practice around machine-learning triage varies widely between centres.
Discussion
Residual confounding cannot be excluded. Strengths include the sample size and blinded outcome assessment.
Implementation will depend on staffing, cost and local protocols. Residual confounding cannot be excluded.
Conclusions
Machine-learning triage should be considered for children with asthma to improve time to diagnosis.
References
1. Novak M, Ibrahim B, et al. Delayed cord clamping and surgical site infection. N Engl J Med. 2010;168(5):407-1439.
2. Ahmed M, Mueller M, et al. Low-dose aspirin and 30-day readmission. Crit Care Med. 2024;123(6):735-1526.
3. Zhang J, Okafor I, et al. Nurse-led discharge planning and surgical site infection. N Engl J Med. 2016;17(10):329-1763.
4. Silva P, Ibrahim F, et al. A sepsis alert system and length of stay. Eur Heart J. 2022;154(6):520-1323.
5. Mensah K, Haddad K, et al. Early enteral nutrition and in-hospital mortality. Ann Intern Med. 2013;44(2):458-1263.
6. Yilmaz G, Zhang L, et al. High-flow nasal oxygen and in-hospital mortality. Lancet. 2011;189(3):181-1541.
7. Rossi U, Kim N, et al. Early enteral nutrition and emergency department revisits. Crit Care Med. 2021;134(7):868-1113.
8. Ahmed H, Haddad K, et al. SGLT2 inhibitors and health-related quality of life. Crit Care Med. 2018;27(11):585-1593.
Appendices
Ethics approval: IRB-6446.
Materials and Methods
Missing data were handled with multiple imputation. Eligible participants were children with asthma. Analyses were adjusted for age, sex and baseline severity.
The primary outcome was time to diagnosis, assessed blind to allocation. Analyses were adjusted for age, sex and baseline severity. Eligible participants were children with asthma.
Results
Effects were consistent across prespecified subgroups. The adjusted odds ratio was 0.90 (95% CI 0.27 to 0.71). 2766 participants were included (median age 80 years). Serious adverse events were rare and similar between groups.
| Group | n | Events | Rate % | OR | 95% CI |
|---|---|---|---|---|---|
| Intervention | 240 | 69 | 6.7 | 0.72 | 0.46–1.36 |
| Control | 819 | 11 | 37.4 | 0.70 | 0.45–1.7 |
| Centre A | 307 | 24 | 31.9 | 0.57 | 0.28–1.7 |
| Age ≥ 65 | 714 | 50 | 2.5 | 0.72 | 0.55–1.22 |
Values are n (%) unless stated otherwise.

Shown for children with asthma; error bars are 95% confidence intervals.
time to diagnosis occurred in 14% of the intervention group versus 27% of controls. Serious adverse events were rare and similar between groups. time to diagnosis occurred in 9% of the intervention group versus 14% of controls.
time to diagnosis occurred in 6% of the intervention group versus 33% of controls. time to diagnosis occurred in 17% of the intervention group versus 29% of controls. time to diagnosis occurred in 20% of the intervention group versus 22% of controls.