European Journal of Trauma and Emergency Surgery

, Volume 43, Issue 6, pp 805–822 | Cite as

Survival prediction of trauma patients: a study on US National Trauma Data Bank

  • I. Sefrioui
  • R. Amadini
  • J. Mauro
  • A. El Fallahi
  • M. Gabbrielli
Original Article



Exceptional circumstances like major incidents or natural disasters may cause a huge number of victims that might not be immediately and simultaneously saved. In these cases it is important to define priorities avoiding to waste time and resources for not savable victims. Trauma and Injury Severity Score (TRISS) methodology is the well-known and standard system usually used by practitioners to predict the survival probability of trauma patients. However, practitioners have noted that the accuracy of TRISS predictions is unacceptable especially for severely injured patients. Thus, alternative methods should be proposed.


In this work we evaluate different approaches for predicting whether a patient will survive or not according to simple and easily measurable observations. We conducted a rigorous, comparative study based on the most important prediction techniques using real clinical data of the US National Trauma Data Bank.


Empirical results show that well-known Machine Learning classifiers can outperform the TRISS methodology. Based on our findings, we can say that the best approach we evaluated is Random Forest: it has the best accuracy, the best area under the curve, and k-statistic, as well as the second-best sensitivity and specificity. It has also a good calibration curve. Furthermore, its performance monotonically increases as the dataset size grows, meaning that it can be very effective to exploit incoming knowledge. Considering the whole dataset, it is always better than TRISS. Finally, we implemented a new tool to compute the survival of victims. This will help medical practitioners to obtain a better accuracy than the TRISS tools.


Random Forests may be a good candidate solution for improving the predictions on survival upon the standard TRISS methodology.


Survival prediction Trauma patients National trauma data bank Classification Machine learning TRISS methodology 



This research work was carried out with the support of the Erasmus Mundus program of the European Union. We would also like to thank the Committee on Trauma, American College of Surgeons for providing us the National Trauma Data Bank used to conduct the experiments.

Compliance with ethical requirements

Conflict of interest

Imane Sefrioui, Roberto Amadini, Jacopo Mauro, Abdellah El Fallahi, and Maurizio Gabbrielli declare that they have no conflict of interest.


  1. 1.
    Gebhart ME, Pence R. START triage: does it work? Disaster Manag Response. 2007;5(3):68–73.CrossRefPubMedGoogle Scholar
  2. 2.
    Koenig KL, Schultz CH. Koenig and Schultz’s Disaster Medicine: Comprehensive Principles and Practices. Cambridge medicine. Cambridge University Press; 2009. Available from:
  3. 3.
    Rehn M, Andersen JE, Vigerust T, Krüger AJ, Lossius HM. A concept for major incident triage: full-scaled simulation feasibility study. BMC Emergency Medicine. 2010;10(1):1–7.
  4. 4.
    van Camp LA, Delooz H. Current trauma scoring systems and their applications: a review. In: Trauma operative procedures. Topics in anaesthesia and critical care. Berlin: Springer; 1999. p. 9–29.Google Scholar
  5. 5.
    Wolfgang FD. Research and uniform reporting. In: Soreide E, Grande CM, editors. Prehospital trauma Care. Boca Raton: CRC Press; 2001. p. 131–152.Google Scholar
  6. 6.
    Sasser S, Varghese M, Kellermann A, Lormand J. Prehospital trauma care systems. World Health Organization; 2005.Google Scholar
  7. 7.
    Linn S, Knoller N, Giligan CG, Dreifus U. The sky is a limit: errors in prehospital diagnosis by flight physicians. Am J Emerg Med. 1997;15(3):316–20.CrossRefPubMedGoogle Scholar
  8. 8.
    Regel G, Seekamp A, Pohlemann T, Schmidt U, Bauer H, Tscherne H. Does the accident patient need to be protected from the emergency doctor? Der Unfallchirurg. 1998;101(3):160–75.CrossRefPubMedGoogle Scholar
  9. 9.
    Boyd CR, Tolson MA, Copes WS. Evaluating trauma care: the TRISS method. Trauma Score and the Injury Severity Score. J Trauma. 1987;27(4):370–8.Google Scholar
  10. 10.
    Breiman L. Random forests. Mach Learn. 2001;45(1):5–32.CrossRefGoogle Scholar
  11. 11.
    TRISS—overview and desktop calculator; 2016.
  12. 12.
    TRISS—scoring systems for ICU and surgical patients; 2016.
  13. 13.
    Kilgo P, Meredith J, Osler T. Incorporating recent advances to make the TRISS approach universally available. J Trauma. 2006;60(5):1002–9.Google Scholar
  14. 14.
    Cox DR. The regression analysis of binary sequences. J R Stat Soc Ser B (Methodological). 1958;20(2): 215–242. Available from:
  15. 15.
    Schetinin V, Jakaite L, Jakaitis J, Krzanowski WJ. Bayesian decision trees for predicting survival of patients: a study on the US National Trauma Data Bank. Comput Methods Progr Biomed. 2013;111(3):602–12.CrossRefGoogle Scholar
  16. 16.
    Rughani AI, Dumont TM, Lu Z, Bongard J, Horgan MA, Penar PL, et al. Use of an artificial neural network to predict head injury outcome. J Neurosurg. 2010;113(3):585–90.CrossRefPubMedGoogle Scholar
  17. 17.
    Mitchell TM. Machine learning. McGraw Hill series in computer science. Maidenheach: McGraw-Hill; 1997.Google Scholar
  18. 18.
    Haykin S. Neural networks: a comprehensive foundation. 2nd ed. Upper Saddle River: Prentice Hall PTR; 1998.Google Scholar
  19. 19.
    Rumelhart DE, Hinton GE, Williams RJ. Learning internal representations by error propagation. In: Parallel distributed processing: explorations in the microstructure of cognition, vol. 1. Massachusetts: MIT Press; 1986. pp. 318–362.Google Scholar
  20. 20.
    Quinlan R. C4.5: Programs for machine learning. San Francisco: Morgan Kaufmann Publishers; 1993.Google Scholar
  21. 21.
    Frank E, Witten IH. Generating accurate rule sets without global optimization. In: ML. San Francisco: Morgan Kaufmann; 1998. pp. 144–151.Google Scholar
  22. 22.
    Altman NS. An introduction to kernel and nearest-neighbor nonparametric regression. Am Stat. 1992;46(3):175–85.Google Scholar
  23. 23.
    le Cessie S, van Houwelingen JC. Ridge estimators in logistic regression. Appl Stat. 1992;41(1):191–201.CrossRefGoogle Scholar
  24. 24.
    Kologlu M, Elker D, Altun H, Sayek I. Validation of MPI and PIA II in two different groups of patients with secondary peritonitis. Hepato-gastroenterology. 2000;48(37):147–51.Google Scholar
  25. 25.
    Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ. Multiple organ dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med. 1995;23(10):1638–52.CrossRefPubMedGoogle Scholar
  26. 26.
    Biondo S, Ramos E, Deiros M, Ragué JM, De Oca J, Moreno P, et al. Prognostic factors for mortality in left colonic peritonitis: a new scoring system. J Am Coll Surg. 2000;191(6):635–42.CrossRefPubMedGoogle Scholar
  27. 27.
    Le Gall JR, Lemeshow S, Saulnier F. A new simplified acute physiology score (SAPS II) based on a European/North American multicenter study. JAMA. 1993;270(24):2957–63.CrossRefPubMedGoogle Scholar
  28. 28.
    John GH, Langley P. Estimating continuous distributions in bayesian classifiers. In: UAI. San Francisco: Morgan Kaufmann; 1995. pp. 338–345.Google Scholar
  29. 29.
    Platt J. Fast training of support vector machines using sequential minimal optimization. In: Advances in kernel methods—support vector learning. Cambridge: MIT Press; 1998.Google Scholar
  30. 30.
    Keerthi SS, Shevade SK, Bhattacharyya C, Murthy KRK. Improvements to Platt’s SMO algorithm for SVM classifier design. Neural Comput. 2001;13(3):637–49.CrossRefGoogle Scholar
  31. 31.
    Hastie T, Tibshirani R. Classification by pairwise coupling. In: NIPS. vol. 10. Cambridge: MIT Press; 1998. pp. 507–513.Google Scholar
  32. 32.
    Wu X, Kumar V, Quinlan JR, Ghosh J, Yang Q, Motoda H, et al. Top 10 algorithms in data mining. Knowl Inf Syst. 2008;14(1):1–37.CrossRefGoogle Scholar
  33. 33.
    Entezari-Maleki R, Rezaei A, Minaei-Bidgoli B. Comparison of classification methods based on the type of attributes and sample size. JCIT. 2009;4(3):94–102.CrossRefGoogle Scholar
  34. 34.
    Committee on Trauma ACoS. NTDB User Manual v. 7.2. 2009.
  35. 35.
    For Prognosis IIM, of clinical trials in TBI A. Injury/disease related events. 2010.
  36. 36.
    Arlot S, Celisse A. A survey of cross-validation procedures for model selection. Stat Surv. 2010;4:40–79.CrossRefGoogle Scholar
  37. 37.
    Fawcett T. ROC graphs: notes and practical considerations for researchers. Palo Alto: HP Laboratories; 2004.Google Scholar
  38. 38.
    Carletta J. Assessing agreement on classification tasks: the kappa statistic. Comput Linguist. 1996;22(2):249–54.Google Scholar
  39. 39.
    Hall M, Frank E, Holmes G, Pfahringer B, Reutemann P, Witten IH. The WEKA data mining software: an update. SIGKDD Explor. Newsl. 2009;11(1):10–8.Google Scholar
  40. 40.
    Cawley GC, Talbot NLC. On over-fitting in model selection and subsequent selection bias in performance evaluation. J Mach Learn Res. 2010;11:2079–107.Google Scholar
  41. 41.
    Platt J. Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods. In: Advances in large margin classifiers. Cambridge: MIT Press; 1999. pp. 61–74.Google Scholar
  42. 42.
    Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3(1):32–5.CrossRefPubMedGoogle Scholar
  43. 43.
    Hosmer DW, Lemeshow S. Applied logistic regression. Applied logistic regression. London: Wiley; 2004. Available from:
  44. 44.
  45. 45.
    Zhu X, Davidson I. Knowledge discovery and data mining: challenges and realities. Hershey: IGI Global; 2007.CrossRefGoogle Scholar
  46. 46.
    Palmer C. Major trauma and the injury severity score—where should we set the bar? Ann Adv Automot Med. 2007;19:13–29.Google Scholar
  47. 47.
    Java runtime environment.
  48. 48.
    Hunter A, Kennedy L, Henry J, Ferguson I. Application of neural networks and sensitivity analysis to improved prediction of trauma survival. Comput Methods Programs Biomed. 2000;62(1):11–9.CrossRefPubMedGoogle Scholar
  49. 49.
    Silva A, Cortez P, Santos MF, Gomes L, Neves J. Mortality assessment in intensive care units via adverse events using artificial neural networks. Artif Intell Med. 2006;36(3):223–34.CrossRefPubMedGoogle Scholar
  50. 50.
    Dc S, As S. A data mining approach for prediction of heart disease using neural networks. IJCET. 2012;3(3):30–40.Google Scholar
  51. 51.
    Larson J, Newman F. An implementation of scatter search to train neural networks for brain lesion recognition. J Math. 2011;4(3):203–11.Google Scholar
  52. 52.
    Kim S, Kim W, Park RW. A comparison of intensive care unit mortality prediction models through the use of data mining techniques. Healthc Inform Res. 2015;17(4):232–43.CrossRefGoogle Scholar
  53. 53.
    Corporation I. SPSS Software; 2013.
  54. 54.
    Michelle S, Hari R, Bryan C, Dua A, Del Junco D, Wade C, et al. Prehospital triage of trauma patients using the random forest computer algorithm. J Surg Res. 2013;187:371–6.Google Scholar
  55. 55.
    Patil BM, Joshi RC, Toshniwal D, Biradar S. A new approach: role of data mining in prediction of survival of burn patients. J Med Syst. 2011;35(6):1531–42.CrossRefPubMedGoogle Scholar
  56. 56.
    Lefering R, Huber-Wagner S, Nienaber U, Maegele M, Bouillon B. Update of the trauma risk adjustment model of the trauma register DGU\(^{\rm TM}\): the revised injury severity classification, version II. Crit Care. 2014;18(5):476. doi: 10.1186/s13054-014-0476-2.
  57. 57.
    Restrepo-Álvarez CA, Valderrama-Molina CO, Giraldo-Ramírez N, Constain-Franco A, Puerta A, León AL, et al. Trauma severity scores. Colombian. J Anesthesiol. 2016;44(4):317–23.Google Scholar
  58. 58.
    Amadini R, Sefrioui I, Mauro J, Gabbrielli M. Fast post-disaster emergency vehicle scheduling. In: DCAI. vol. 217 of Advances in intelligent systems and computing. Berlin: Springer; 2013. pp. 219–226.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2017

Authors and Affiliations

  1. 1.Faculty of Sciences of TetouanUniversity Abdelmalek EssaadiTétouanMorocco
  2. 2.Department of Computing and Information SystemsThe University of MelbourneMelbourneAustralia
  3. 3.Department of InformaticsUniversity of OsloOsloNorway
  4. 4.National School of Applied Sciences of TetouanUniversity Abdelmalek EssaadiTétouanMorocco
  5. 5.Department of Computer ScienceUniversity of Bologna/Lab. Focus INRIABolognaItaly

Personalised recommendations