Answers are grounded in this paper and cite their sources.
Maritime safety is essential for preventing vessel accidents, ensuring safe navigation, and protecting marine ecosystems, particularly in increasingly congested waterways. Existing maritime risk assessment studies have demonstrated the usefulness of AIS-based trajectory prediction, traffic-pattern analysis, and hazard identification; however, many approaches remain focused on individual risk factors and provide limited integration of vessel behaviour, environmental conditions, and operational context within adaptive and real-time risk assessment frameworks. To address this gap, this study develops a predictive framework for proactive maritime risk evaluation using Automatic Identification System (AIS) data integrated with environmental, traffic, and behavioural factors. The proposed approach combines robust data cleaning, normalization, missing-value handling, categorical and binary encoding with autoencoder-based feature extraction and Principal Component Analysis (PCA) for compact and informative representations. Subsequently, a hybrid Uniform Manifold Approximation and Projection (UMAP)–Self-Supervised Learning (SSL)–Deep Embedded Clustering (DEC) framework is employed to learn risk-related representations and identify distinct maritime risk patterns. Compared with the evaluated UMAP and SSL approaches, the proposed DEC-based approach achieves superior performance, with 91.33% accuracy, 92.37% F1-score, a 0.8312 silhouette score, and a 0.31 Davies–Bouldin Index. Overall, these findings demonstrate the effectiveness of the proposed framework in distinguishing maritime risk patterns and support its potential for proactive, real-time maritime safety monitoring and decision support.
No takes yet. Share an insight, caveat, or question.
Nigam et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: