Two-dimensional ETTC-labeled longitudinal and lateral conflicts for interpretable real-time crash risk prediction in freeway interchange diverging areas.
DOI: 10.1371/journal.pone.0344623
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This study addresses the limitations of existing real-time crash prediction models (RTCPMs) in freeway interchange diverging areas, which typically rely on macroscopic traffic parameters that fail to capture microscopic vehicle interactions. The research aims to develop an interpretable, conflict-type-aware framework that distinguishes between longitudinal (car-following) and lateral (lane-changing) crash risks. This distinction is critical because diverging areas involve frequent lane changes, creating complex interaction mechanisms that differ from standard freeway segments. The methodology involved collecting high-resolution trajectory data from 12 interchange diverging areas along two mountainous freeways in China. Drone video footage was recorded over 212 hours during peak traffic periods. Using a YOLOX and DeepSORT-based computer vision framework, the researchers extracted trajectories for 9,876 vehicles, deriving 48 microscopic motion parameters and surrogate safety measures (SSMs). The study innovatively applied Extended Time-to-Collision (ETTC), a two-dimensional metric, to identify lateral conflicts, while using traditional metrics for longitudinal conflicts. This data was used to label traffic states at 30-second intervals. Four machine learning models—Random Forest, Neural Network, Support Vector Machine, and XGBoost—were trained to predict crash risks. The SHAP (SHapley Additive exPlanations) framework was then employed to interpret the models and identify key risk factors. The results demonstrated that the XGBoost model achieved optimal predictive performance. Analysis revealed that lateral conflicts exhibited longer durations and higher crash risks than longitudinal conflicts, with severe conflicts concentrated within 200 meters upstream of exit ramps. The interpretability analysis identified Modified Time-to-Collision (MTTC), which incorporates relative acceleration, alongside Stopping Headway Distance and Time-to-Collision, as the most decisive factors for both conflict types. These SSMs ranked highest in predictive contribution, highlighting the importance of dynamic kinematic indicators over static traffic flow parameters. The significance of this work lies in its provision of a scientific foundation for proactive traffic safety management. By explicitly modeling lateral lane-changing risks and identifying specific spatial and temporal risk drivers, the study supports the design of targeted dangerous driving warning systems. The findings enable traffic operators to implement precise interventions in interchange diverging areas, moving beyond generic macroscopic monitoring to address the specific microscopic behaviors that lead to crashes. This approach enhances the interpretability and effectiveness of real-time crash prediction in complex freeway environments.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | PubMed Central | — | — | 1 | 2026-06-20 |
| archive | success | unpaywall | — | — | 2 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-20 |
| chunk | success | chunk | — | — | 1 | 2026-06-20 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-20 |
| enrich | success | openalex | — | — | 1 | 2026-06-20 |
| promote | success | — | — | — | 1 | 2026-06-20 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: crash risk outcomes