Ambulance optimization study shows simulated gains with restrictive travel and service assumptions

Engage the EMS director, dispatch supervisor, medical director, fleet planners and labor representatives around uneven coverage. Ask whether existing redeployment rules, hospital delays or unavailable crews dominate response time, and whether reliable incident and vehicle-state histories exist. Offer a bounded historical replay and operating-readiness assessment. The value hypothesis is improved positioning under the agency's actual constraints, subject to independent comparison with present practice. Qualify whether the department can fund data preparation and frontline review. Do not turn simulated percentage improvements into promised minutes saved, clinical benefit or fewer required ambulances. Applicability will be limited where travel estimates or unit availability cannot be reconstructed.

Prototype an advisory optimization service outside the critical dispatch path. Prerequisites include time-aligned CAD events, unit status, routable roads, hospital destination rules and authorized clinical priority logic. Replace simplified travel assumptions with measured local distributions and include turnout, handover, shift changes and simultaneous incidents. Secure event feeds and preserve the state behind each recommendation. A useful proof of value compares median and tail response time by priority and neighborhood, coverage breaches and computation delay with existing rules. Test missing telemetry and service failure. The model should abstain when its inputs are stale; no live assignment authority should follow from a replay result alone.

Dispatch operations should own implementation with a clinical safety lead, GIS analyst, data engineer and crew representatives. Build a baseline, document exclusions, validate the simulator with actual journeys and review recommendations with users before shadow operation. Dependencies include usable CAD exports and agreement on crew and hospital constraints. Governance checkpoints should approve data access, test design, shadow results and every rule change. Proposed acceptance requires no invalid unit assignments in the agreed scenario suite, demonstrated fallback, prespecified priority-specific performance bounds and recorded override reasons. Monitor geographic disparities and repositioning workload. These are proposed criteria, not results; a favorable mean must not conceal worse critical calls.