A simulation based digital twin approach to assessing the organization of response to emergency calls

Abstract

In emergency situations, timely contact with emergency medical communication centers (EMCCs) is critical for patient outcomes. Increasing call volumes and economic constraints are challenging many countries, necessitating organizational changes in EMCCs. This study uses a simulation-based digital twin approach, creating a virtual model of EMCC operations to assess the impact of different organizational scenarios on accessibility. Specifically, we explore two decompartmentalized scenarios where traditionally isolated call centers are reorganized to enable more flexible call distribution. The primary measure of accessibility was service quality within 30 s of call reception. Our results show that decompartmentalization improves service quality by 17% to 21%. This study demonstrates that reducing regional isolation in EMCCs can enhance performance and accessibility with a simulation-based digital twin approach providing a clear and objective method to quantify the benefits."

Fig. 1. Overview of the variability of QoS_30i on one day (07/17/2021) for scenario #1 (As-Is: Partitioned).

It illustrates an example for a single day of the week (17/07/2021) of the variation in quality of service for each slot (QoS_30i) over the 336 30-min slots in the week (from ±17% to ±39%). The target value of 99% was not achieved but in some time slots no incoming call was answered in less than 30 s.

Fig. 2. Overview of the variability of QoS_30i on one day (07/17/2021) for scenario #2 (Interconnected).

It highlights a remarkable result: the variability of QoS_30i is considerably reduced compared with the reference scenario. Even if the target of 99% QoS30 is not achieved in EMCC-A, the values are significantly improved (+11% to +32%) compared with scenario 1.

Fig. 3. Overview of the variability of QoS_30i on one day (07/17/2021) for scenario #3 (Virtualized).

It presents the effect of the virtualized scenario. With this organization, the variability of QoS_30i is improved compared with the existing organization, but is less than with the interconnected organization.

Fig. 4. Population distribution in the Pays de la Loire area.

It presents the distribution of the population in the Pays de la Loire area and the population density for each department. The figure shows that the region is representative of France, with some areas very densely populated and others with lower densities.

Fig. 5. Overview of incoming call volume during the summer period (Monday, May 31 to Sunday, August 15, 2021).

It provides an overview of the volume of incoming calls during the summer period (Monday, May 31 to Sunday, August 15, 2021) and identifies the week of highest activity considered for this study (Monday, July 12 to Sunday, July 18).

Fig. 6. Data preparation example: distribution of call durations.

It shows the frequency distribution of call duration (min and max). It is used as an input to randomly simulate the duration of each call answered.

Fig. 7. Knowledge model of the front-office process.

Knowledge model of the process of receiving calls by an emergency call operator (ECO) that is universally deployed in emergency call centers. Incoming calls arrive in the ECO queue. Some callers leave the queue before being connected to an ECO, in two categories: (1) abandoned calls leave the queue within 10 s and (2) lost calls leave the queue after 10 s. In addition to abandoned and lost calls, calls in the queue are handled by the first available ECO on a first-in, first-out (FIFO) basis. The regulation process focused strictly on the front-office process.

Fig. 8. Overview of the Digital Twin developed with Witness.

It provides an overview of the generic digital twin front-office process of a call center with its input data (on the left) and output responses (on the right: QoS_{_}30_i for each 30-min slot i, and the average value AQoS₃₀) according to the organizational scenario to be tested. Three main input data that are extracted from the real production database are: (1) the number of incoming calls in each 30-min slot (top-left diagram); (2) the number of operators ECO scheduled in each 30-min slot (middle-left diagram); (3) the call duration distribution. The digital model (in the center of Fig. 8) is powered by these input data during a simulation run on the 7 days of the studied week. At the end of a simulation run, two main output data are collected: (1) the value of QoS_30i in each 30-min slot enabling analysis of the variability of the QoS in the week (orange curve on the top-right diagram of Fig. 8). To quantify this variability, the coefficient of variation (CoV % = 100 * Standard Deviation / Mean) is calculated. The larger the CoV, the more QoS_30i is spread out; (2) The average value of QoS_30i is calculated in the 336 30-min slots of the week (bottom-right equation in Fig. 8). This figure is enlarged and split into five figures in Supplementary Figs. 2 to 7.

Fig. 9. Comparison between real data (blue line) and simulated data (orange line).

It shows the comparison of real and simulated values of AQoS₃₀ for the EMCC-A in the existing organization (partitioned). The mean signed deviation between real and simulation data was equal to 7% for this example and the simulated line (orange) logically follows the same developments as the real line (blue). We considered that the behavior of the digital twin was close to the behavior of its real-world twin.

Fig. 10. Scenario 1: Example of two partitioned EMCCs (EMCC-A and EMCC-B).

It shows the partitioned organization, which means that there is no interoperability between the EMCCs. It forces each EMCC to handle only incoming calls routed from its own geographical territory. This scenario corresponds to the existing organization.

Fig. 11. Scenario 2: Example of two interconnected EMCCs (EMCC-A and EMCC-B).

It illustrates the interconnected organization. Each EMCC retains its own queue of calls arriving from its geographical territory. However, an ECO can, if available and if there are no calls in his or her own queue, pick up calls from the queue of another EMCC.

Fig. 12. Scenario 3: Example of two virtualized EMCC (EMCC-A and EMCC-B).

It shows the virtualized organization. In this case, it is with a single queue common to all centers. A call in this queue is picked up on a first-in-first-out (FIFO) basis by the first available ECO from any EMCC.