1 / 21

# Minimizing Average Response Times in a Dynamic Ambulance Management Model

Minimizing Average Response Times in a Dynamic Ambulance Management Model. Thije van Barneveld, CWI, Amsterdam. Region. Equidistant graph Blue: Demand locations Yellow: Hospitals Red: Additional nodes. Ambulance Phases. “Life Cycle” of a Request. State components. State components.

Télécharger la présentation

## Minimizing Average Response Times in a Dynamic Ambulance Management Model

E N D

### Presentation Transcript

1. Minimizing Average Response Times in a Dynamic Ambulance Management Model Thije van Barneveld, CWI, Amsterdam

2. Region • Equidistant graph • Blue: Demand locations • Yellow: Hospitals • Red: Additional nodes

3. Ambulance Phases

4. “Life Cycle” of a Request

5. State components

6. State components • Elapsed service time of ambulances in phase 4 • Destinations and remaining driving times of phase 3 ambulances

7. Actions • Dispatch nearest ambulance • Change in ambulance configuration

8. Objective • Cost in a state: number of patients waiting • Minimize costs: minimize the average number of patients waiting • Minimize the average response time Cost 0 Cost 1 Cost 0

9. Heuristic Solution - Idea • Observe state • Consider all actions • Consider possible scenarios • Combine each action with each scenario • Classify each action and optimize

10. Scenarios • Possible next state • One new request • Ambulances that finish service

11. Scenarios • Possible next state • One new request • Ambulances that finish service

12. Scenarios • Possible next state • One new request • Ambulances that finish service

13. Scenarios • Possible next state • One new request • Ambulances that finish service

14. Scenarios • Possible next state • One new request • Ambulances that finish service

15. Eligible ambulances Eligibleforrespondingto new request: • Nearestidleunassigned ambulance • Nearest busy ambulance at hospital • Nearest busy ambulance on scene, notrequiredto transport Expectedshortest response time

16. Example Classify action: • Scenario probabilityExpectedshortest response time torequest • Sum over scenarios • Take best classified action

17. Results • ’s estimatedusinghistorical data • No hospitals • 4 ambulances

18. Results

19. Results • ’s as before • 2 hospitals: • 6 ambulances

20. Results

21. Questions?

More Related