Enhanced Crash Risk Estimation in Urban Environments: Integrating Multi-Source Data and Advanced Modeling Approaches for the City of Pittsburgh
-
2026-07-31
Details
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
- Crash injury research
- Emergency medical services
- Injury severity
- Transportation corridors
- Urban transportation
- Emergency communication systems
- Data collection
- Center lines
- Literature reviews
- High Injury Network (HIN)
- Emergency medical services dispatch data
- Crash under-reporting
- Corridor-level network screening
-
Resource Type:
-
Geographical Coverage:
-
Corporate Publisher:
-
Description:High Injury Networks help inform the decision which streets an agency treats first, and they are almost always built from police-reported crashes alone. That record is known to miss injuries, and to miss crashes unevenly in space. We test whether a second, independently collected source names different high injury streets. Using City of Pittsburgh data for 2020 to 2024, we merge the street centerline into 4,332 corridors covering 933.5 miles, build a KABCO-aligned severity index for 14,298 emergency medical dispatches from crew-recorded disposition, transport mode and patient-handling fields corrected by a negation-aware reading of the scene text, and assign each of the 18,267 crashes and each dispatch to a single corridor. We then apply the City's rule identically to each layer, flagging a corridor when any 1,500 ft window holds five or more fatal-and-injury events or any 200 ft window holds two or more serious-or-fatal events. EMS attends fewer traffic events than police but records more injuries, 6,529 against 5,429. The two layers select overlapping but distinct networks: 128 crash corridors and 164 EMS corridors, agreeing on 100 of those corridors, which carry 73.0 percent of flagged mileage. The 64 EMS-only corridors are mostly Local and Collector streets and mark candidate under-reported segments. The procedure requires no personal identifiers and no record-level match across data sources.
-
Format:
-
Funding:
-
Download URL:
-
File Type:
-
Collection(s):
-
Main Document Checksum:urn:sha-512:99bbad0528f98ee832f220606d51dd2f412b9d3b16ae19c4b110becea1422b24b24865e8b824b05aff25ca4bacd60e92764b13a485b45ea4b3f8b99293bb2eac