Early Prediction of Massive Transfusion for Patients With Traumatic Hemorrhage: Development of a Multivariable Machine Learning Model.

Objective: Develop a novel machine learning (ML) model to rapidly identify trauma patients with severe hemorrhage at risk of early mortality. Background: The critical administration threshold (CAT, 3 or more units of red blood cells in a 60-minute period) indicates severe hemorrhage and predicts mortality, whereas early identification of such patients improves survival. Methods: Patients from the PRospective, Observational, Multicenter, Major Trauma Transfusion and Pragmatic, Randomized Optimal Platelet, and Plasma Ratio studies were identified as either CAT+ or CAT-. Candidate variables were separated into 4 tiers based on the anticipated time of availability during the patient's assessment. ML models were created with the stepwise addition of variables and compared with the baseline performance of the assessment of blood consumption (ABC) score for CAT+ prediction using a cross-validated training set and a hold-out validation test set. Results: Of 1245 PRospective, Observational, Multicenter, Major Trauma Transfusion and 680 Pragmatic, Randomized Optimal Platelet and Plasma Ratio study patients, 1312 were included in this analysis, including 862 CAT+ and 450 CAT-. A CatBoost gradient-boosted decision tree model performed best. Using only variables available prehospital or on initial assessment (Tier 1), the ML model performed superior to the ABC score in predicting CAT+ patients [area under the receiver-operator curve (AUC = 0.71 vs 0.62)]. Model discrimination increased with the addition of Tier 2 (AUC = 0.75), Tier 3 (AUC = 0.77), and Tier 4 (AUC = 0.81) variables. Conclusions: A dynamic ML model reliably identified CAT+ trauma patients with data available within minutes of trauma center arrival, and the quality of the prediction improved as more patient-level data became available. Such an approach can optimize the accuracy and timeliness of massive transfusion protocol activation.

Mathematical Modeling, In-Human Evaluation and Analysis of Volume Kinetics and Kidney Function After Burn Injury and Resuscitation.

OBJECTIVE: Existing burn resuscitation protocols exhibit a large variability in treatment efficacy. Hence, they must be further optimized based on comprehensive knowledge of burn pathophysiology. A physics-based mathematical model that can replicate physiological responses in diverse burn patients can serve as an attractive basis to perform non-clinical testing of burn resuscitation protocols and to expand knowledge on burn pathophysiology. We intend to develop, optimize, validate, and analyze a mathematical model to replicate physiological responses in burn patients. METHODS: Using clinical datasets collected from 233 burn patients receiving burn resuscitation, we developed and validated a mathematical model applicable to computer-aided in-human burn resuscitation trial and knowledge expansion. Using the validated mathematical model, we examined possible physiological mechanisms responsible for the cohort-dependent differences in burn pathophysiology between younger versus older patients, female versus male patients, and patients with versus without inhalational injury. RESULTS: We demonstrated that the mathematical model can replicate physiological responses in burn patients associated with wide demographic characteristics and injury severity, and that an increased inflammatory response to injury may be a key contributing factor in increasing the mortality risk of older patients and patients with inhalation injury via an increase in the fluid retention. CONCLUSION: We developed and validated a physiologically plausible mathematical model of volume kinetic and kidney function after burn injury and resuscitation suited to in-human application. SIGNIFICANCE: The mathematical model may provide an attractive platform to conduct non-clinical testing of burn resuscitation protocols and test new hypotheses on burn pathophysiology.

Optimizing a decision support system for damage-control resuscitation using mixed methods human factors analysis.

BACKGROUND: Damage-control resuscitation (DCR) improves trauma survival; however, consistent adherence to DCR principles through multiple phases of care has proven challenging. Clinical decision support may improve adherence to DCR principles. In this study, we designed and evaluated a DCR decision support system using an iterative development and human factors testing approach. METHODS: The phases of analysis included initial needs assessment and prototype design (Phase 0), testing in a multidimensional simulation (Phase 1), and testing during initial clinical use (Phase 2). Phase 1 and Phase 2 included hands-on use of the decision support system in the trauma bay, operating room, and intensive care unit. Participants included trauma surgeons, trauma fellows, anesthesia providers, and trauma bay and intensive care unit nurses who provided both qualitative and quantitative feedback on the initial prototype and all subsequent iterations. RESULTS: In Phase 0, 14 (87.5%) of 16 participants noted that they would use the decisions support system in a clinical setting. Twenty-four trauma team members then participated in simulated resuscitations with decision support where 178 (78.1%) of 228 of tasks were passed and 27 (11.8%) were passed with difficulty. Twenty-three (95.8%) completed a postsimulation survey. Following iterative improvements in system design, Phase 2 evaluation included 21 trauma team members during multiple real-world trauma resuscitations. Of these, 15 (71.4%) completed a formal postresuscitation survey. Device-level feedback on a Likert scale (range, 0-4) confirmed overall ease of use (median score, 4; interquartile range, 4-4) and indicated the system integrated well into their workflow (median score, 3; interquartile range, 2-4). Final refinements were then completed in preparation for a pilot clinical study using the decision support system. CONCLUSIONS: An iterative development and human factors testing approach resulted in a clinically useable DCR decision support system. Further analysis will determine its applicability in military and civilian trauma care. LEVEL OF EVIDENCE: Therapeutic/Care Management, Level V.

Mathematical model of volume kinetics and renal function after burn injury and resuscitation.

This paper presents a mathematical model of blood volume kinetics and renal function in response to burn injury and resuscitation, which is applicable to the development and non-clinical testing of burn resuscitation protocols and algorithms. Prior mathematical models of burn injury and resuscitation are not ideally suited to such applications due to their limited credibility in predicting blood volume and urinary output observed in wide-ranging burn patients as well as in incorporating contemporary knowledge of burn pathophysiology. Our mathematical model consists of an established multi-compartmental model of blood volume kinetics, a hybrid mechanistic-phenomenological model of renal function, and novel lumped-parameter models of burn-induced perturbations in volume kinetics and renal function equipped with contemporary knowledge on burn-related physiology and pathophysiology. Using the dataset collected from 16 sheep, we showed that our mathematical model can be characterized with physiologically plausible parameter values to accurately predict blood volume kinetic and renal function responses to burn injury and resuscitation on an individual basis against a wide range of pathophysiological variability. Pending validation in humans, our mathematical model may serve as an effective basis for in-depth understanding of complex burn-induced volume kinetic and renal function responses as well as development and non-clinical testing of burn resuscitation protocols and algorithms.