EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Explainable Machine Learning-Based Maritime Crew Health Risk Assessment for Proactive Fatigue Monitoring and Decision Support

Thu Phuong Nguyen, Nghia Chung, Duc Chuan Nguyen, Phuoc Quy Phong Nguyen, Van Quy Nguyen

International Journal on Advanced Science, Engineering and Information Technology · 22.06.2026 · 16 (3) · 824-836

DOI: 10.18517/ijaseit.16.3.13813 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal on Advanced Science, Engineering and Information Technology, online erschienen am 22.06.2026. Verfasst von Thu Phuong Nguyen und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study used Machine Learning and Explainable Artificial Intelligence to predict fatigue and assess health risk. We built a complete dataset of 500 maritime crew records, including demographic, physiological, psychological, occupational, and operational factors. The proposed Maritime Crew Health Risk Index (MCHRI) normalization combines stress level, mental stress score, workload index, sea state, sleep quality, sleep duration, and noise exposure into a single normalized risk score ranging from 0 to 100. The performance of four ensemble learning algorithms for predicting fatigue was assessed: Random Forest, XGBoost, LightGBM, and CatBoost. With an R² of 0.968 and RMSE of 2.942, CatBoost had the highest predictive ability compared to LightGBM (R² = 0.951), XGBoost (R² = 0.946), and Random Forest (R² = 0.895). The results of the goodness-of-fit analysis also validated CatBoost's performance, with R² = 0.9668 and MAE = 2.3533. The risk stratification showed that 17.8% were low risk, 61.4% moderate risk, and 20.8% high risk. The top five fatigue determinants found by SHAP-based explainability were sleep quality, health-risk category, stress level, sleep time, and workload index. Results also indicated that fatigue levels rose from ~46 to ~66 units as workloads increased, and that voyage duration was weakly related to fatigue. The proposed framework would provide clear and practical means of decision support for proactive crew health monitoring, fatigue mitigation, and maritime safety management.

Zitieren

APA

Nguyen, T. P., Chung, N., Nguyen, D. C., Nguyen, P. Q. P., & Nguyen, V. Q. (2026). Explainable Machine Learning-Based Maritime Crew Health Risk Assessment for Proactive Fatigue Monitoring and Decision Support. International Journal on Advanced Science, Engineering and Information Technology, 16(3), 824-836. https://doi.org/10.18517/ijaseit.16.3.13813

Vancouver

Nguyen TP, Chung N, Nguyen DC, Nguyen PQP, Nguyen VQ. Explainable Machine Learning-Based Maritime Crew Health Risk Assessment for Proactive Fatigue Monitoring and Decision Support. International Journal on Advanced Science, Engineering and Information Technology. 2026;16(3):824-836. doi:10.18517/ijaseit.16.3.13813

BibTeX
@article{Nguyen2026,
  author = {Nguyen, Thu Phuong and Chung, Nghia and Nguyen, Duc Chuan and Nguyen, Phuoc Quy Phong and Nguyen, Van Quy},
  title = {Explainable Machine Learning-Based Maritime Crew Health Risk Assessment for Proactive Fatigue Monitoring and Decision Support},
  journal = {International Journal on Advanced Science, Engineering and Information Technology},
  year = {2026},
  volume = {16},
  number = {3},
  pages = {824--836},
  doi = {10.18517/ijaseit.16.3.13813}
}
Kostenlos starten →