Integrating survival analysis and machine learning for modeling overtaking duration on rural two-lane highways
- 1Yunnan Key Laboratory of Digital Communications, Yunnan Communications Investment & Construction Group Co., Ltd, Kunming, China
- 2National Engineering Research Center for Efficient Maintenance, Safety and Durability of Roads and Bridges, Broadvision Engineering Consultants Co., Ltd, Kunming, China
Abstract
Overtaking duration is closely associated with overtaking safety because longer maneuvers increase exposure in the opposing traffic lane. However, evidence on overtaking duration on rural mountainous two-lane highways remains limited, particularly under mixed-traffic conditions. This study investigates the determinants of overtaking duration on rural mountainous two-lane highways using an explainable survival machine-learning framework. Unmanned aerial vehicle videos were collected, and vehicle trajectories were extracted to obtain detailed overtaking maneuver data, including complete and right-censored observations. To model censored duration data and capture nonlinear relationships between overtaking duration and its influencing factors, a random survival forest (RSF) model was developed and compared with a conventional log-logistic accelerated failure time (AFT) model. The results show the average values of overtaking duration and overtaking distance were 10.2 seconds and 201.2 meters, respectively. The RSF model achieved a higher C-index than the log-logistic AFT model, indicating better discrimination of overtaking completion times. SHAP results identified the initial speed of the overtaking vehicle, lateral distance, initial speed difference, and overtaking vehicle type as the most influential factors, with pronounced nonlinear effects observed for overtaking-vehicle speed and speed difference. These findings highlight the potential of interpretable survival machine learning for analyzing overtaking-duration data and provide insights into overtaking behavior and safety on rural mountainous two-lane highways.
Related
- Parsing altered brain connectivity in neurodevelopmental disorders by integrating graph-based normative modeling and deep generative networks
- Multiscale predictive cellular modeling: integrating hypothesis grammars, digital twins, and multi-omics for In silico oncology and precision theranostics
- Modeling the effects of stress and premature hive exiting behavior on honey bee (Apis mellifera) colony survival
- Hepatocellular carcinoma screening in patients with compensated hepatitis C virus (HCV)-related cirrhosis aware of their HCV status improves survival: A modeling approach
- Integrating Spatially Adjusted Protein Summaries for Survival Prediction in Spatial Proteomics
BibTeX
@article{shi2026integrating,
title = {{Integrating survival analysis and machine learning for modeling overtaking duration on rural two-lane highways}},
author = {Chuanhe Shi and Shuai Fu and Yuanchang Zhao and Jing He and Peng Ye and Wenchen Yang},
journal = {PLOS One},
year = {2026},
month = sep,
doi = {10.1371/journal.pone.0357092},
url = {https://doi.org/10.1371/journal.pone.0357092},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Integrating survival analysis and machine learning for modeling overtaking duration on rural two-lane highways
AU - Chuanhe Shi
AU - Shuai Fu
AU - Yuanchang Zhao
AU - Jing He
AU - Peng Ye
AU - Wenchen Yang
PY - 2026
DA - 2026/09/01
JO - PLOS One
DO - 10.1371/journal.pone.0357092
UR - https://doi.org/10.1371/journal.pone.0357092
AB - Overtaking duration is closely associated with overtaking safety because longer maneuvers increase exposure in the opposing traffic lane. However, evidence on overtaking duration on rural mountainous two-lane highways remains limited, particularly under mixed-traffic conditions. This study investigates the determinants of overtaking duration on rural mountainous two-lane highways using an explainable survival machine-learning framework. Unmanned aerial vehicle videos were collected, and vehicle trajectories were extracted to obtain detailed overtaking maneuver data, including complete and right-censored observations. To model censored duration data and capture nonlinear relationships between overtaking duration and its influencing factors, a random survival forest (RSF) model was developed and compared with a conventional log-logistic accelerated failure time (AFT) model. The results show the average values of overtaking duration and overtaking distance were 10.2 seconds and 201.2 meters, respectively. The RSF model achieved a higher C-index than the log-logistic AFT model, indicating better discrimination of overtaking completion times. SHAP results identified the initial speed of the overtaking vehicle, lateral distance, initial speed difference, and overtaking vehicle type as the most influential factors, with pronounced nonlinear effects observed for overtaking-vehicle speed and speed difference. These findings highlight the potential of interpretable survival machine learning for analyzing overtaking-duration data and provide insights into overtaking behavior and safety on rural mountainous two-lane highways.
SN - 1932-6203
ER -