Development of a field artificial intelligence triage tool: Confidence in the prediction of shock, transfusion, and definitive surgical therapy in patients with truncal gunshot wounds.
Adult
Artificial Intelligence
Blood Transfusion
/ statistics & numerical data
Emergency Medical Services
/ methods
Feasibility Studies
Female
Hemorrhage
/ epidemiology
Humans
Injury Severity Score
Male
Models, Cardiovascular
ROC Curve
Retrospective Studies
Risk Assessment
/ methods
Shock
/ epidemiology
Thoracic Injuries
/ complications
Trauma Centers
Triage
/ methods
Wounds, Gunshot
/ complications
Young Adult
Journal
The journal of trauma and acute care surgery
ISSN: 2163-0763
Titre abrégé: J Trauma Acute Care Surg
Pays: United States
ID NLM: 101570622
Informations de publication
Date de publication:
01 06 2021
01 06 2021
Historique:
entrez:
21
5
2021
pubmed:
22
5
2021
medline:
24
9
2021
Statut:
ppublish
Résumé
In-field triage tools for trauma patients are limited by availability of information, linear risk classification, and a lack of confidence reporting. We therefore set out to develop and test a machine learning algorithm that can overcome these limitations by accurately and confidently making predictions to support in-field triage in the first hours after traumatic injury. Using an American College of Surgeons Trauma Quality Improvement Program-derived database of truncal and junctional gunshot wound (GSW) patients (aged 16-60 years), we trained an information-aware Dirichlet deep neural network (field artificial intelligence triage). Using supervised training, field artificial intelligence triage was trained to predict shock and the need for major hemorrhage control procedures or early massive transfusion (MT) using GSW anatomical locations, vital signs, and patient information available in the field. In parallel, a confidence model was developed to predict the true-class probability (scale of 0-1), indicating the likelihood that the prediction made was correct, based on the values and interconnectivity of input variables. A total of 29,816 patients met all the inclusion criteria. Shock, major surgery, and early MT were identified in 13.0%, 22.4%, and 6.3% of the included patients, respectively. Field artificial intelligence triage achieved mean areas under the receiver operating characteristic curve of 0.89, 0.86, and 0.82 for prediction of shock, early MT, and major surgery, respectively, for 80/20 train-test splits over 1,000 epochs. Mean predicted true-class probability for errors/correct predictions was 0.25/0.87 for shock, 0.30/0.81 for MT, and 0.24/0.69 for major surgery. Field artificial intelligence triage accurately identifies potential shock in truncal GSW patients and predicts their need for MT and major surgery, with a high degree of certainty. The presented model is an important proof of concept. Future iterations will use an expansion of databases to refine and validate the model, further adding to its potential to improve triage in the field, both in civilian and military settings. Prognostic, Level III.
Sections du résumé
BACKGROUND
In-field triage tools for trauma patients are limited by availability of information, linear risk classification, and a lack of confidence reporting. We therefore set out to develop and test a machine learning algorithm that can overcome these limitations by accurately and confidently making predictions to support in-field triage in the first hours after traumatic injury.
METHODS
Using an American College of Surgeons Trauma Quality Improvement Program-derived database of truncal and junctional gunshot wound (GSW) patients (aged 16-60 years), we trained an information-aware Dirichlet deep neural network (field artificial intelligence triage). Using supervised training, field artificial intelligence triage was trained to predict shock and the need for major hemorrhage control procedures or early massive transfusion (MT) using GSW anatomical locations, vital signs, and patient information available in the field. In parallel, a confidence model was developed to predict the true-class probability (scale of 0-1), indicating the likelihood that the prediction made was correct, based on the values and interconnectivity of input variables.
RESULTS
A total of 29,816 patients met all the inclusion criteria. Shock, major surgery, and early MT were identified in 13.0%, 22.4%, and 6.3% of the included patients, respectively. Field artificial intelligence triage achieved mean areas under the receiver operating characteristic curve of 0.89, 0.86, and 0.82 for prediction of shock, early MT, and major surgery, respectively, for 80/20 train-test splits over 1,000 epochs. Mean predicted true-class probability for errors/correct predictions was 0.25/0.87 for shock, 0.30/0.81 for MT, and 0.24/0.69 for major surgery.
CONCLUSION
Field artificial intelligence triage accurately identifies potential shock in truncal GSW patients and predicts their need for MT and major surgery, with a high degree of certainty. The presented model is an important proof of concept. Future iterations will use an expansion of databases to refine and validate the model, further adding to its potential to improve triage in the field, both in civilian and military settings.
LEVEL OF EVIDENCE
Prognostic, Level III.
Identifiants
pubmed: 34016929
doi: 10.1097/TA.0000000000003155
pii: 01586154-202106000-00019
doi:
Types de publication
Journal Article
Research Support, U.S. Gov't, Non-P.H.S.
Langues
eng
Sous-ensembles de citation
IM
Pagination
1054-1060Informations de copyright
Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.
Références
Benson M, Koenig KL, Schultz CH. Disaster triage: START, then SAVE—a new method of dynamic triage for victims of a catastrophic earthquake. Prehosp Disaster Med . 1996;11(2):117–124.
Sasser SM, Hunt RC, Faul M, Sugerman D, Pearson WS, Dulski T, Wald MM, Jurkovich GJ, Newgard CD, Lerner EB; Centers for Disease Control and Prevention (CDC). Guidelines for field triage of injured patients: recommendations of the National Expert Panel on Field Triage, 2011. MMWR Recomm Rep . 2012;61(RR-1):1–20.
Sewalt CA, Venema E, Wiegers EJA, Lecky FE, Schuit SCE, den Hartog D, Steyerberg EW, Lingsma HF. Trauma models to identify major trauma and mortality in the prehospital setting. Br J Surg . 2020;107(4):373–380.
Deeb AP, Phelos HM, Peitzman AB, Billiar TR, Sperry JL, Brown JB. Disparities in rural versus urban field triage: risk and mitigating factors for undertriage. J Trauma Acute Care Surg . 2020;89:246–253.
Schellenberg M, Benjamin E, Bardes JM, Inaba K, Demetriades D. Undertriaged trauma patients: who are we missing? J Trauma Acute Care Surg . 2019;87(4):865–869.
Uribe-Leitz T, Jarman MP, Sturgeon DJ, Harlow AF, Lipsitz SR, Cooper Z, Salim A, Newgard CD, Haider AH. National study of triage and access to trauma centers for older adults. Ann Emerg Med . 2020;75(2):125–135.
Augustin P, Guivarch E, Pellenc Q, Castier Y, Boudinet S, Tanaka S, Montravers P, Tran-Dinh A. Performance of pre-hospital evaluations in ruling out invasive chest stab wounds. Scand J Trauma Resusc Emerg Med . 2020;28(1):33.
Voskens FJ, van Rein EAJ, van der Sluijs R, et al. Accuracy of prehospital triage in selecting severely injured trauma patients. JAMA Surg . 2018;153(4):322–327.
Haas B, Stukel TA, Gomez D, Zagorski B, De Mestral C, Sharma SV, Rubenfeld GD, Nathens AB. The mortality benefit of direct trauma center transport in a regional trauma system: a population-based analysis. J Trauma Acute Care Surg . 2012;72(6):1510–1515; discussion 1515-7.
Newgard CD, Yang Z, Nishijima D, et al; Western Emergency Services Translational Research Network Investigators. Cost-effectiveness of field trauma triage among injured adults served by emergency medical services. J Am Coll Surg . 2016;222(6):1125–1137.
Bertsimas D, Dunn J, Steele DW, Trikalinos TA, Wang Y. Comparison of machine learning optimal classification trees with the pediatric emergency care applied research network head trauma decision rules. JAMA Pediatr . 2019;173(7):648–656.
Bertsimas D, Dunn J, Velmahos GC, Kaafarani HMA. Surgical risk is not linear: derivation and validation of a novel, user-friendly, and machine-learning-based Predictive OpTimal Trees in Emergency Surgery Risk (POTTER) Calculator. Ann Surg . 2018;268(4):574–583.
Duceau B, Alsac JM, Bellenfant F, et al. Prehospital triage of acute aortic syndrome using a machine learning algorithm. Br J Surg . 2020;107:995–1003.
Fernandes M, Mendes R, Vieira SM, Leite F, Palos C, Johnson A, Finkelstein S, Horng S, Celi LA. Predicting intensive care unit admission among patients presenting to the emergency department using machine learning and natural language processing. PLoS One . 2020;15(3):e0229331.
Collins GS, Reitsma JB, Altman DG, Moons KG. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ . 2015;350:g7594.
Yücel N, Lefering R, Maegele M, Vorweg M, Tjardes T, Ruchholtz S, Neugebauer EA, Wappler F, Bouillon B, Rixen D; Polytrauma Study Group of the German Trauma Society. Trauma Associated Severe Hemorrhage (TASH)-Score: probability of mass transfusion as surrogate for life threatening hemorrhage after multiple trauma. J Trauma . 2006;60(6):1228–1236; discussion 1236-7.
Purcell GP. What makes a good clinical decision support system. BMJ . 2005;330(7494):740–741.
Kawamoto K, Houlihan CA, Balas EA, Lobach DF. Improving clinical practice using clinical decision support systems: a systematic review of trials to identify features critical to success. BMJ . 2005;330(7494):765.
Hanna K, Harris C, Trust MD, Bernard A, Brown C, Hamidi M, Joseph B. Multicenter validation of the Revised Assessment of Bleeding and Transfusion (RABT) score for predicting massive transfusion. World J Surg . 2020;44(6):1807–1816.
McLaughlin DF, Niles SE, Salinas J, Perkins JG, Cox ED, Wade CE, Holcomb JB. A predictive model for massive transfusion in combat casualty patients. J Trauma . 2008;64(Suppl 2):S57–S63; discussion S63.
El-Menyar A, Abdelrahman H, Al-Thani H, Mekkodathil A, Singh R, Rizoli S. The FASILA score: a novel bio-clinical score to predict massive blood transfusion in patients with abdominal trauma. World J Surg . 2020;44(4):1126–1136.
Cotton BA, Dossett LA, Haut ER, Shafi S, Nunez TC, Au BK, Zaydfudim V, Johnston M, Arbogast P, Young PP. Multicenter validation of a simplified score to predict massive transfusion in trauma. J Trauma . 2010;69(Suppl 1):S33–S39.
Nunez TC, Voskresensky IV, Dossett LA, Shinall R, Dutton WD, Cotton BA. Early prediction of massive transfusion in trauma: simple as ABC (assessment of blood consumption)? J Trauma . 2009;66(2):346–352.
Schreiber MA, Perkins J, Kiraly L, Underwood S, Wade C, Holcomb JB. Early predictors of massive transfusion in combat casualties. J Am Coll Surg . 2007;205(4):541–545.
Mutschler M, Nienaber U, Münzberg M, Wölfl C, Schoechl H, Paffrath T, Bouillon B, Maegele M; TraumaRegister DGU. The Shock Index revisited — a fast guide to transfusion requirement? A retrospective analysis on 21,853 patients derived from the TraumaRegister DGU. Crit Care . 2013;17(4):R172.
Brockamp T, Nienaber U, Mutschler M, Wafaisade A, Peiniger S, Lefering R, Bouillon B, Maegele M; TraumaRegister DGU. Predicting on-going hemorrhage and transfusion requirement after severe trauma: a validation of six scoring systems and algorithms on the TraumaRegister DGU. Crit Care . 2012;16(4):R129.
Joseph B, Khan M, Truitt M, Jehan F, Kulvatunyou N, Azim A, Jain A, Zeeshan M, Tang A, O’Keeffe T. Massive transfusion: the Revised Assessment of Bleeding and Transfusion (RABT) score. World J Surg . 2018;42(11):3560–3567.