New York City's 911 ambulance system, the largest municipal EMS operation in the world, is getting help from an unlikely source: a team of computer scientists in Los Angeles building a virtual copy of the entire network.

Researchers at the University of Southern California are partnering with the FDNY on an NSF-funded project to develop what they call a "probabilistic digital shadow" of NYC's EMS operations. The idea is to create a data-driven simulation of the city's ambulance system, accurate enough that planners can test new dispatch and deployment strategies inside the model before trying anything on real streets and real patients.

The stakes are concrete. FDNY EMS handles roughly 1.5 million medical calls a year with around 4,000 EMTs and paramedics, a workforce stretched by attrition and surging demand that has never fully receded since the COVID-19 pandemic. Average response times for life-threatening emergencies have hovered around 7 to 10 minutes citywide, with longer waits concentrated in lower-income neighborhoods in Brooklyn, the Bronx, and Queens. A 2022 city comptroller audit found those times had worsened, exceeding official targets.

The current dispatch system relies largely on fixed rules and static posting plans, approaches designed for predictable conditions that the modern city rarely provides. Traffic alone creates radically different challenges borough to borough: Manhattan gridlock bears almost no resemblance to the highway-dependent response geometry of eastern Queens.

The USC research team aims to replace guesswork with machine learning models that can forecast call volumes, estimate ambulance travel times in real time, and account for rare but catastrophic surges like the roughly 6,500 calls FDNY received on a single day in March 2020, nearly double normal volume. The simulation would let FDNY test how different staffing patterns, staging locations, or dispatch rules would perform across a wide range of scenarios before anyone commits to changing actual operations.

The $210,743 award covers USC's portion of the project, funded through the NSF's Smart and Connected Communities program, which requires a direct government partner. Additional subawards are planned, suggesting NYC-based collaborators will be embedded in the work. Total project funding across all partners is likely larger, as SCC grants typically range well above this figure.

Whether AI-driven dispatch changes would face resistance from FDNY's EMS unions is an open question. Labor groups, represented by FDNY EMS Local 2507, have long pushed for higher pay and more staffing as the core fix for response time problems, and any move toward algorithmic decision-making would likely draw scrutiny.

For now, the research phase is just beginning. Whether the virtual city eventually reshapes how the real one deploys its ambulances depends on what the simulations show, and whether FDNY leadership chooses to act on them.