Study III — Infection presentations: dispatch, phenotype, and high-risk triage
Study III examines concordance across system classifications—dispatch reason for call, on-scene phenotype (ESS), and high-risk triage (RETTS)—and
uses interpretable exploratory modelling to characterise heterogeneity in routinely captured data.
Core interpretation principle
Because ESS contributes to RETTS, the study quantifies concordance/alignment across stages rather than independent predictive validity.
Key results (study-level)
- On-scene infection-coded ESS labels were present in ~64% of “Fever/Infection” calls, indicating moderate EMCC–ESS concordance.
- A gradient-boosting model discriminated high-risk triage (RETTS Red/Orange) with moderate accuracy (AUC reported in the manuscript).
- Most predictive signals included on-scene ESS code keywords, the response-time interval, and patient age; PD showed a threshold rise in risk beyond ~60 years and non-linear contextual patterns.
- The apparent decline in predicted high-risk probability with longer response times is interpreted as consistent with preferential early response to the most critical cases (selection/priority effects), not as evidence that delay is beneficial.
Status / availability
Manuscript. Public link will be added when shareable.
Adaptive EMS Allocation 