Description: | A. METHODOLOGY | A1. INJURY AND FATALITY DATA SOURCESThis dataset is sourced from traffic crash data from the San Francisco Police Department (SFPD), hospital data from Zuckerberg San Francisco General Hospital (ZSFG), ambulance operator data from American Medical Response (AMR), and the San Francisco Fire Department Emergency Medical Services (EMS). These data are linked using a non-probabilistic deterministic method to match records between these three data sources. Once they are matched, any details about the crash circumstances are defaulted to come from SFPD and the final injury severity determination is sourced from ZSFG Hospital. The dataset also includes un-matched SFPD severe and fatal crash victims that had associated geolocations, as well as un-linked ZSFG Hospital data from the Trauma Registry that had linked geolocation data from the ambulance operators’ pick- up locations.A2. MAPPING INJURIES AND FATALITESOnly injuries and deaths with valid geographic information are mapped. All geocodable injury data is represented on the simplified San Francisco street centerline model maintained by the Department of Public Works (SFDPW). Data on transportation-related injuries on streets in San Francisco reported to SFPD, or treated at ZSFG, is queried and aggregated on a yearly basis, and is current up to December 2024. Injuries and deaths occurring at complex intersections with multiple roadways are mapped onto a single point, and injuries and fatalities occurring on highways are excluded.A3. CALCULATING KILLED AND SEVERELY INJURED PER MILESan Francisco’s street centerline data layer is corridorized into segments that are at least 0.25 miles long along segments that share the same street name. Then, the point data of Killed or Severely Injured (KSI) is plotted onto the map and a KSI per mile per corridor is calculated based on a count of points that geographically intersect a given corridor.A4. CONSTRUCTING THE NETWORKThe base map for the HIN is created by filtering corridors based on minimum length. Any corridor that is less than 0.25 miles in total length is excluded from eligibility onto the network. Then, corridors are selected based on a minimum cutpoint of greater than or equal to 10 KSI per mile per corridor.A5. REMOVING STUBS FROM NETWORKAdditional processing of the map includes finding street segments of the base HIN map that are greater than 500 feet away from the last on-HIN KSI on that street. These “stubs” are removed before producing the final HIN map to further prioritize city resources on the streets with the highest 5-year incidence of severe injuries and fatalities.| B. UPDATE FREQUENCY | This dataset will not be updated once published. However, SFDPH, SFMTA, and SFPD reserve the right to update this map if injury data is added, removed or revised.| C. OTHER CRITICAL INFORMATION | The 2024 HIN represents a snapshot in time (2020-2024) where severe and fatal injuries are most concentrated. It may not reflect current conditions or changes to the City’s transportation system. Although prior incidents can be indicative of future incidents, the 2024 HIN is not a prediction (probability) of future risk. The HIN approach contrasts with risk-based analysis, which focuses on locations determined to be more dangerous with increased risk or danger often calculated by dividing the number of injuries or collisions by vehicle volumes to estimate risk of injury per vehicle. The HIN provides information regarding the streets where injuries, particularly severe and fatal, are concentrated in San Francisco based on injury counts; it is not an assessment of whether a street or location is dangerous. The 2024 High Injury Network is derived from the more severe injury outcomes (count of severe and fatal injuries) and may not cover locations with high numbers of less severe injury collisions.Hospital and emergency medical service records from which SFPD-unreported injury and reclassified injury collisions are derived are protected by the Health Insurance Portability and Accountability Act and state medical privacy laws, thus have strict confidentiality and privacy requirements. As of May 2026, SFDPH is working in conjunction with the Office of Compliance and Privacy Affairs, Zuckerberg San Francisco General Hospital (ZSFG), and the SFMTA to ensure that the data SFDPH shares is in compliance with federal and state privacy laws. Intersection and other small-area specific counts of severe and fatal injuries are intentionally excluded from this document in compliance with those laws.Recommended Citation San Francisco Population Health Division Center for Data Science. 2026. San Francisco High Injury Network: 2024 Update. San Francisco, CA.Related Datasets Traffic Crashes Resulting in InjuryTraffic Crashes Resulting in Injury: Parties InvolvedTraffic Crashes Resulting in Injury: Victims InvolvedTraffic Crashes Resulting in Fatality2022 High Injury Network2017 High Injury Network
Copyright Text: | A. CUSTODIAN |
Geodata custodian: Center for Data Science, cds.phd@sfdph.org, Population Health Division, Department of Public Health
| B. CREATED | 20260323, Center for Data Science