Assessing the impact of external factors on the occurrence of emergencies

TL;DR

The paper tackles how external factors influence hourly emergency occurrences to inform EMS relocation. It combines classical statistics (correlation, Chi-squared, t-tests, information value) with ML interpretability (SHAP, permutation importance) on XGBoost and MLP, using CHUV data from 2015–2021 across 11 stations, and benchmarks against a simple hour-only baseline. The key finding is that the hour of the day is the dominant predictor, with other factors offering limited predictive value, largely due to correlation with time; all models deliver similar performance to the hour-only baseline. Practically, this suggests focusing resources on time-of-day patterns and considering stochastic approaches, such as an inhomogeneous Poisson process, for EMS relocation rather than relying on complex ML models.

Abstract

This study investigates the impact of 19 external factors, related to weather, road traffic conditions, air quality, and time, on the occurrence of emergencies using historical data provided by the dispatch center of the Centre Hospitalier Universitaire Vaudois (CHUV). This center is responsible for managing Emergency Medical Service (EMS) resources in the majority of the French-speaking part of Switzerland. First, classical statistical methods, such as correlation, Chi-squared test, Student's $t$-test, and information value, are employed to identify dependencies between the occurrence of emergencies and the considered parameters. Additionally, SHapley Additive exPlanations (SHAP) values and permutation importance are computed using eXtreme Gradient Boosting (XGBoost) and Multilayer Perceptron (MLP) models. The results indicate that the hour of the day, along with correlated parameters, plays a crucial role in the occurrence of emergencies. Conversely, other factors do not significantly influence emergency occurrences. Subsequently, a simplified model that considers only the hour of the day is compared with our XGBoost and MLP models. These comparisons reveal no significant difference between the three models in terms of performance, supporting the use of the basic model in this context. These observations provide valuable insights for EMS resource relocation strategies, benefit predictive modeling efforts, and inform decision-making in the context of EMS. The implications extend to enhancing EMS quality, making this research essential.

Figures (103)

• Figure 1: Region of interest.
• Figure 2: Ambulances stations location. Correspondence between station number and name can be found in Table \ref{['stations stats']}.
• Figure 3: Location of all different considered stations and traffic points. Ambulance stations are in red, weather stations in blue, counting points in black, and air quality stations in orange.
• Figure 4: Number of emergencies related to Aigle's station regarding different time intervals.
• Figure 5: Correlation matrix, Morges.