Original Articles Open access

Early versus late mortality among fatal trauma cases: a comparative analysis using statistical methods and machine learning algorithms

Authors

Published
1 September 2026
Section
Original Articles
DOI
10.4081/ecj.2026.14775

Abstract

Early mortality after trauma is a major concern in emergency care. This study aimed to compare deaths occurring within the first 24 hours with those occurring later using statistical analysis and supervised machine learning, and to assess the effect of class imbalance on model performances. This retrospective study included all consecutive trauma deaths recorded between May and August 2021 in the hospital information system of a tertiary care training and research hospital. The cohort was restricted to patients who died during hospitalization because the aim was to distinguish mortality timing among fatal trauma cases rather than to predict mortality versus survival. Mortality time was classified as death within 24 hours or death after 24 hours from emergency department admission. Clinical, laboratory, and injury-related variables were analyzed using univariate tests. Multiple supervised machine learning models were trained with a stratified 80:20 train/test split and evaluated by Area Under the Curve (AUC). Five resampling techniques were employed to address class imbalance. Hemoglobin and hematocrit differed significantly between early and late mortality groups (p<0.05), and injury patterns were associated with mortality timing. On the imbalanced dataset, the support vector machine (SVM) achieved the highest AUC (0.800). After resampling, Gaussian Naïve Bayes (GNB) combined with ADASYN achieved the highest AUC (0.847). These findings suggest that mortality timing among fatal trauma cases can be explored using supervised machine learning; however, the retrospective single-center design, modest sample size, class imbalance, and exclusion of survivors limit direct applicability to real-time emergency department triage. The model should therefore be interpreted as an exploratory mortality- timing classifier rather than as a general trauma mortality prediction tool.

Keywords

Ethics Approval

Ethical approval was obtained from the Clinical Research Ethics Committee of the University of Health Sciences Kocaeli Training and Research Hospital (approval number: 2021/138; dated December 23, 2021).

Supporting Agencies

This research did not receive any funding.

Data Availability Statement

Due to privacy and confidentiality concerns, the data are not publicly available. However, they can be made available upon reasonable request to the corresponding author, subject to the completion of a signed data access agreement.

References

  1. [1] Wang H, Robinson RD, Moore B, et al. Predictors of early versus late mortality in pelvic trauma patients. Scand J Trauma Resusc Emerg Med 2016;24:27. DOI: https://doi.org/10.1186/s13049-016-0220-9
  2. [2] Jin WYY, Jeong JH, Kim DH, et al. Factors predicting the early mortality of trauma patients. Ulus Travma Acil Cerrahi Derg 2018;24:532-8. DOI: https://doi.org/10.5505/tjtes.2018.29434
  3. [3] Naemi A, Schmidt T, Mansourvar M, et al. Machine learning techniques for mortality prediction in emergency departments: a systematic review. BMJ Open 2021;11:e052663. DOI: https://doi.org/10.1136/bmjopen-2021-052663
  4. [4] Zhang T, Nikouline A, Lightfoot D, Nolan B. Machine learning in the prediction of trauma outcomes: a systematic review. Ann Emerg Med 2022;80:440-55. DOI: https://doi.org/10.1016/j.annemergmed.2022.05.011
  5. [5] Collins GS, Moons KGM, Dhiman P, et al. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ 2024;385:q902. DOI: https://doi.org/10.1136/bmj.q902
  6. [6] Sobrino J, Shafi S. Timing and causes of death after injuries. Proc (Bayl Univ Med Cent) 2013;26:120-3. DOI: https://doi.org/10.1080/08998280.2013.11928934
  7. [7] Pfeifer R, Tarkin IS, Rocos B, Pape HC. Patterns of mortality and causes of death in polytrauma patients-has anything changed? Injury 2009;40:907-911. DOI: https://doi.org/10.1016/j.injury.2009.05.006
  8. [8] Kauvar DS, Lefering R, Wade CE. Impact of hemorrhage on trauma outcome. J Trauma 2006;60:S3-S11. DOI: https://doi.org/10.1097/01.ta.0000199961.02677.19
  9. [9] Rady MY, Smithline HA, Blake H, Nowak R, Rivers E. Shock index vs conventional vital signs. Ann Emerg Med 1994;24:685-90. DOI: https://doi.org/10.1016/S0196-0644(94)70279-9
  10. [10] Steyerberg EW. Clinical prediction models: a practical approach to development, validation, and updating. Springer; 2019. DOI: https://doi.org/10.1007/978-3-030-16399-0
  11. [11] Gravesteijn BY, Nieboer D, Ercole A, et al. Machine learning vs regression in TBI prognostication. J Clin Epidemiol 2020;122:95-107. DOI: https://doi.org/10.1016/j.jclinepi.2020.03.005
  12. [12] Christodoulou E, Ma J, Collins GS, et al. No performance benefit of ML over logistic regression. BMJ 2019;366:l5398.
  13. [13] He H, Garcia EA. Learning from imbalanced data. IEEE Trans Knowl Data Eng 2009;21:1263-84. DOI: https://doi.org/10.1109/TKDE.2008.239
  14. [14] Branco P, Torgo L, Ribeiro RP. Predictive modeling on imbalanced domains. ACM Comput Surv 2016;49:31. DOI: https://doi.org/10.1145/2907070
  15. [15] Chawla NV, Bowyer KW, Hall LO, Kegelmeyer WP. SMOTE. J Artif Intell Res 2002;16:321-57. DOI: https://doi.org/10.1613/jair.953
  16. [16] He H, Bai Y, Garcia EA, Li S. ADASYN. In: Proc Int Joint Conf Neural Netw 2008:1322-1328.
  17. [17] Lemaitre G, Nogueira F, Aridas CK. Imbalanced-Learn: A Python toolbox to tackle the curse of imbalanced datasets in machine learning. J Machine Learning Res 2017;18:1–5.
  18. [18] Hyland SL, Faltys M, Hüser M, et al. Early prediction of circulatory failure using ML. Nat Med 2020;26:364-73. DOI: https://doi.org/10.1038/s41591-020-0789-4
  19. [19] Lundberg SM, Lee, SI. A unified approach to interpreting model predictions. Adv Neural Informat Processing Systems 2017;30. https://proceedings.neurips.cc/paper_files/paper/2017/file/8a20a8621978632d76c43dfd28b67767-Paper.pdf.
  20. [20] Rau CS, Wu SC, Chuang JF, et al. ML survival prediction in trauma. J Clin Med 2019;8:799. DOI: https://doi.org/10.3390/jcm8060799

Metrics

Citations and online attention from Dimensions, Altmetric and PlumX.

PlumX Metrics

License

Copyright (c) 2026 the Author(s)

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

PAGEPress has chosen to apply the Creative Commons Attribution NonCommercial 4.0 International License (CC BY-NC 4.0) to all manuscripts to be published.