Enhancing the Golden Hour: Prehospital Identification of TBI and Relevant Endophenotypes

Tuesday, March 24, 2026

8:45 AM - 9:45 AM Central Time

Location: Connections Central - RST 09

NB

Neeraj Badjatia, MD, MS, FCCM

University of Maryland Medical Center, R. Adams Cowley Shock Trauma Center, Maryland

Disclosure(s): No relevant financial relationship(s) to disclose.
MW

Matthew Woodward, DO

Dr
Univ of Maryland, Maryland

Disclosure information not submitted.
SY

Shiming Yang, PhD

Dr
Univ of Maryland, Maryland

Disclosure information not submitted.
KD

Ken Dalton, MD

Dr
University of Maryland Medical Center, R. Adams Cowley Shock Trauma Center

Disclosure information not submitted.
rf

ryan felix, N/A

Mr
Univ of Maryland, Maryland

Disclosure information not submitted.
JP

Jamie Podell, MD

Dr
University of Maryland Medical Center, R. Adams Cowley Shock Trauma Center

Disclosure information not submitted.
LC

Lujie Chen, PhD

Dr
University of Maryland, Baltimore County

Disclosure information not submitted.
GS

Gary Schwartzbauer, MD, PhD

Dr
University of Maryland Medical Center, R. Adams Cowley Shock Trauma Center

Disclosure information not submitted.
WT

William Teeter, MD

Dr
Univ of Maryland, Maryland

Disclosure information not submitted.
DF

Douglas Floccare, MD

Dr
Univ of Maryland, Maryland

Disclosure information not submitted.
PH

Peter Hu, PhD

Dr
University of Maryland School of Medicine

Disclosure information not submitted.

Introduction: During emergency transport, clinical assessment and vital signs may lack the sensitivity to identify traumatic brain injury (TBI) and identify specific TBI endophenotypes of shock, coagulopathy, and polytrauma which may have important implications for appropriate triaging and timely delivery of life-saving interventions. We evaluated the ability of machine learning (ML) algorithms to identify the presence of TBI and specific endophenotypes of shock, coagulopathy and polytrauma during air transport to a trauma center.

Methods: We identified a cohort of consecutive trauma patients aged 18-65 transported from the scene of injury via helicopter to an urban academic trauma center and collected prehospital clinical data and continuous vital signs. TBI was defined and stratified by severity using the head AIS score. We used ElasticNet (regularized regression) and XGBoost (gradient boosting), comparing three variable sets: clinical variables only, continuous physiologic monitoring data only, and a combined clinical and physiological data, to develop four predictive models: (1) mild vs. moderate-severe TBI, (2) presence/absence of polytrauma in moderate-severe TBI, (3) presence/absence of coagulopathy in TBI, and (4) presence/absence of shock in TBI,.

Results: 1,025 patients (median age 38, IQR: 27-53; 70% male) were identified. Median Glasgow Coma Scale score was 15 (IQR: 13-15). Across all predictive models, ML algorithms exhibited good predictive discrimination, with area under the receiver operator curve (AUROC) of 0.79 (0.75-0.84), 0.89 (0.85-0.92), 0.81 (0.72-0.89), and 0.86 (0.82-0.91) for TBI severity, polytrauma, coagulopathy, and shock, respectively. Clinical data best predicted TBI severity and polytrauma, while physiologic data improved prediction of shock and coagulopathy.

Conclusions: Machine learning algorithms integrating clinical and continuous physiological monitoring can improve identification of TBI and its endophenotypes during prehospital transport.