RoadTrace’s Whitepaper
Our latest study reveals how RoadTrace’s data-driven approach predicts high-risk locations faster and more accurately, helping authorities prevent accidents before they happen.
The Study: Comparing Two Methods
This research, conducted in South-Eastern England, compared:
A traditional model based on five years of historical crash data
A new predictive model using only harsh-braking clusters from connected vehicles
The goal: to determine which method detects risk areas earlier and with higher precision.
3× more predictive information
Compared with five years of historical crash records, harsh-braking clusters provided 22 % usable prediction signals, versus only 6 % from KSI (Killed or Seriously Injured) data.
Accuracy improves by 22 % overall
When both datasets were tested on the same network, the connected-vehicle model achieved a 22 % higher prediction rate for identifying future collision sites.
Much faster detection
The connected-vehicle approach enabled much faster detection of emerging collision zones, well before accidents occurred.
Hidden danger zones become visible
The model even predicted collisions that traditional crash data never identified, despite five full years of records.
Get the Full White Paper
Learn how connected-vehicle data transforms crash prevention strategies and supports smarter infrastructure planning.
The full study includes detailed methodology, data visualisations, and validation results.