Abstract
Accurate and effective diagnosis of actual injury severity can be problematic in trauma patients. Inherent physiologic compensatory mechanisms may prevent accurate diagnosis and mask true severity in many circumstances. The objective of this project was the development and validation of a multiparameter machine learning algorithm and system capable of predicting the need for life-saving interventions (LSIs) in trauma patients. Statistics based on means, slopes, and maxima of various vital sign measurements corresponding to 79 trauma patient records generated over 110,000 feature sets, which were used to develop, train, and implement the system. Comparisons among several machine learning models proved that a multilayer perceptron would best implement the algorithm in a hybrid system consisting of a machine learning component and basic detection rules. Additionally, 295,994 feature sets from 82 h of trauma patient data showed that the system can obtain 89.8 % accuracy within 5 min of recorded LSIs. Use of machine learning technologies combined with basic detection rules provides a potential approach for accurately assessing the need for LSIs in trauma patients. The performance of this system demonstrates that machine learning technology can be implemented in a real-time fashion and potentially used in a critical care environment.
Similar content being viewed by others
References
Alpaydin E (2004) Introduction to machine learning, 1st edn. MIT Press, Cambridge, pp 1–16
Batchinsky AI, Salinas J, Jones JA, Necsoiu C, Cancio LC (2009) Predicting the need to perform life-saving interventions in trauma patients using new vital signs and artificial neural networks. Lect Notes Comput Sc 5651:390–394
Baxt WG, Jones G, Fortlage D (1990) The trauma triage rule: a new, resource-based approach to the prehospital identification of major trauma victims. Ann Emerg Med 19:1401–1406
Buchanan BG (2005) A (very) brief history of artificial intelligence. AI Magazine 26:53–60
Chen L, Reisner AT, Gribok A, Reifman J (2009) Exploration of prehospital vital sign trends for the prediction of trauma outcomes. Prehosp Emerg Care 13:286–294
Duda RO, Hart PE, Stork DG (2001) Pattern classification, 2nd edn. John Wiley and Sons, New York, pp 215–349
Garner A, Lee A, Harrison K, Schultz CH (2001) Comparative analysis of multiple-casualty incident triage algorithms. Ann Emerg Med 38:541–548
Hall M, Frank E, Holmes G, Pfahringer B, Reutemann P, Witten IH (2009) The WEKA data mining software: an update. SIGKDD Explor 11:10–18
Holcomb JB, Salinas J, McManus JM, Miller CC, Cooke WH, Convertino VA (2005) Manual vital signs reliably predict need for life-saving interventions in trauma patients. J Trauma 59:821–829
Hravnak M, Devita MA, Clontz A, Edwards L, Valenta C, Pinsky MR (2008) Cardiorespiratory instability before and after implementing an integrated monitoring system. Crit Care Med 177:A842
Hravnak M, Edwards L, Clontz A, Valenta C, DeVita MA, Pinsky MR (2008) Defining the incidence of cardio-respiratory instability in step-down unit patients using an electronic integrated monitoring system. Arch Intern Med 168:1300–1308
Patela VL, Shortliffea EH, Stefanellic M, Szolovitsd P, Bertholde MR, Bellazzic R, Abu-Hanna A (2009) The coming of age of artificial intelligence in medicine. Artif Intell Med 46:5–17
Rohwer R, Wynne-Jones M, Wysotzki F (1994) Neural Networks. In: Michie D, Spiegelhalter DJ, Taylor CC (eds) Machine learning, neural and statistical classification. Series in artificial intelligence. Ellis Horwood Publishing, New York, pp 84–105
Tarassenko L, Hann A, Young D (2006) Integrated monitoring and analysis for early warning of patient deterioration. Br J Anaesth 97:64–68
Acknowledgments
This work was supported by the National Trauma Institute, the US Army Combat Casualty Care Research Program, and the State of Texas Emergency Technology Fund. We acknowledge the expertise, dedication, and professionalism of the Emergency Medical Services paramedics, nurses, and staff in Houston; Denise Hinds, Timothy Welch, and Jeannette Podbielski (the University of Texas Health Science Center in Houston, Texas, USA); and Kevin Stitcher (Athena GTX, Inc). We also thank Athena GTX, Inc. for the use of the Murphy factor to support algorithm development.
Conflict of interest
None.
Author information
Authors and Affiliations
Corresponding author
Additional information
Disclaimer: The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense.
Rights and permissions
About this article
Cite this article
Liu, N.T., Holcomb, J.B., Wade, C.E. et al. Development and validation of a machine learning algorithm and hybrid system to predict the need for life-saving interventions in trauma patients. Med Biol Eng Comput 52, 193–203 (2014). https://doi.org/10.1007/s11517-013-1130-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11517-013-1130-x