Responsible analysis of EMS systems
The research combines register-based cohort studies, qualitative interviews, emergency medical dispatch data, triage information, response-time analysis, and interpretable analytical methods.
System understanding before automation
The programme combines qualitative evidence about dispatch work with observational analyses of large-scale EMS data. The methodological objective is system understanding under uncertainty, especially distributions, heterogeneity, queue dynamics, and nonlinear patterns.
Register-based cohort studies
Retrospective observational analyses of routinely collected emergency medical communication centre and EMS data from Stockholm, Sweden, covering 2017–2022.
Qualitative interviews
Inductive qualitative content analysis of emergency medical dispatchers’ experiences of prioritisation, queue governance, resource stewardship, and ambulance scarcity.
Interpretability-first modelling
Machine learning is used to explore nonlinear conditional patterns using tools such as partial dependence and individual conditional expectation plots.
What the data can and cannot show
Routine EMS data can reveal important system patterns. They also require careful interpretation because response time, triage, and dispatch priority are produced inside the system being studied.
Exploration, not deployment
Machine learning in this research is used to support system understanding. It is not presented as a ready-to-use dispatch automation tool, triage replacement, or operational optimisation product.
Observational inference
Analyses describe patterns in real-world system data and require careful interpretation of confounding, selection, endogenous allocation, and system-generated exposure.
Model transparency
Discrimination metrics are reported as transparency descriptors, while the central interpretive focus is on explanation profiles and conditional patterns.
Governance before deployment
Any operational translation requires prospective evaluation, human-factors design, explicit governance, and safeguards against unintended consequences.
Responsible use of routine EMS data
The research uses routinely collected operational data to understand EMS system behaviour under uncertainty. Findings should be interpreted as system-level evidence, not as patient-level advice or validated real-time decision support.
Public presentation of data is limited by confidentiality, privacy, ethical approval, and public-sector information governance. Patient-level records, granular geography, and operationally sensitive details are not disclosed.
