Using Deep Learning in Severity Analysis of At-Fault Motorcycle Rider Crashes
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 critical safety issue of motorcycle crash severity, motivated by the disproportionately high fatality rates among motorcyclists compared to passenger vehicle occupants. In the United States, motorcyclist fatalities occur 27 times more frequently per vehicle mile traveled than those of car occupants, with 4,976 fatalities recorded in 2014. The authors argue that conventional statistical models, such as multinomial logit and probit models, often suffer from poor prediction accuracy and reliance on restrictive assumptions. To overcome these limitations, the paper introduces a deep learning framework named "DeepScooter" to predict injury severity outcomes for at-fault motorcycle riders. The research utilized five years (2010–2014) of police-reported crash data from Louisiana, focusing exclusively on crashes where the motorcycle rider was at fault to isolate rider-specific factors. The final dataset comprised 6,853 crashes, characterized by 16 predictor variables including roadway geometry, lighting, weather, rider condition, and collision type. Injury severity was categorized using the KABCO scale: fatal, incapacitating, non-incapacitating, possible/complaint, and no injury. The authors developed the DeepScooter model using the R with H2O.ai platform, employing a multi-step process involving data splitting (50% training, 25% validation, 25% testing), initial model development, and hyperparameter tuning. The final model used an adaptive learning algorithm with 100 epochs, 128 hidden layers, and early stopping criteria to optimize performance. The results demonstrated that the DeepScooter framework significantly outperformed traditional statistical methods. The final model achieved 100% accuracy on the training data and 94% accuracy on the test data, with the lowest misclassification rates among the tested configurations. The analysis identified specific factors strongly associated with higher crash severity intensity. These included rider ejection, crashes occurring on two-way roadways without physical separation, curved road alignments, and incidents happening on weekends. Additionally, the study noted that while most crashes occurred during daylight and clear weather, these conditions were linked to higher speeds and thus more severe outcomes. The significance of this work lies in its demonstration that deep learning can provide superior predictive accuracy for crash severity compared to conventional statistical inference methods. By identifying key risk factors such as road geometry and rider ejection, the findings offer actionable insights for improving motorcycle safety infrastructure and policies. The DeepScooter framework serves as a novel tool for transportation safety analysis, suggesting that algorithmic modeling can effectively handle the complex, high-dimensional nature of crash data to better predict and potentially mitigate severe injuries.
Key finding
The DeepScooter deep learning framework predicted motorcycle crash severity with 94% accuracy on test data, identifying rider ejection, two-way roads without separation, curved alignments, and weekend occurrences as significant contributors to severe outcomes.
Methodology
modeling
Sample size: 6853
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. Discovered via author_sweep_intake on 2026-05-28.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-28 |
| archive | success | canonical_url | — | — | 7 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | success | — | — | — | 1 | 2026-05-28 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- motorcyclist skill
- motorcycle crash typology
- vru crash typology
- helmet protective
- motorcycle conspicuity
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