Methodological overview
The programme combines qualitative evidence about real-time dispatch work with observational analyses of large-scale EMS data. The methodological objective is system understanding under uncertainty—especially distributions, heterogeneity, and non-linear patterns—rather than automation.
Interpretation guardrail: Where machine learning is used, it is exploratory and interpretability-first. Performance metrics are contextual only and must not be interpreted as evidence for deployable dispatch or triage automation.
Data (high level)
- Routinely collected EMS operational metadata (timestamps, classifications, response-time intervals)
- On-scene clinical classifications where applicable (e.g., ESS and triage labels)
- Reported in aggregate with privacy-by-design principles
Analytic principles
- Distribution-first: response-time distributions and tail delays
- Heterogeneity: patterns vary by context and vulnerability
- Queue dynamics: prioritisation shapes observed associations
- Interpretability: explanation and measurement over prediction-for-deployment
Adaptive EMS Allocation 