Related Experiment Video
Updated: Dec 6, 2025
Systems Analysis of the Neuroinflammatory and Hemodynamic Response to Traumatic Brain Injury
Published on: May 27, 2022
Multi-Biomarker Prediction Models for Multiple Infection Episodes Following Blunt Trauma
Amy Tsurumi1,2,3, Patrick J Flaherty4, Yok-Ai Que5
1Department of Surgery, Massachusetts General Hospital, and Harvard Medical School, 50 Blossom St., Their 340, Boston, MA 02114, USA.
Abstract:
Severe trauma predisposes patients to multiple independent infection episodes (MIIEs), leading to augmented morbidity and mortality. We developed a method to identify increased MIIE risk before clinical signs appear, which is fundamentally different from existing approaches entailing infections' detection after their establishment. Applying machine learning algorithms to genome-wide transcriptome data from 128 adult blunt trauma patients' (42 MIIE cases and 85 non-cases) leukocytes collected ≤48 hr of injury and ≥3 days before any infection, we constructed a 15-transcript and a 26-transcript multi-biomarker panel model with the least absolute shrinkage and selection operator (LASSO) and Elastic Net, respectively, which accurately predicted MIIE (Area Under Receiver Operating Characteristics Curve [AUROC] [95% confidence intervals, CI]: 0.90 [0.84-0.96] and 0.92 [0.86-0.96]) and significantly outperformed clinical models. Gene Ontology and network analyses found various pathways to be relevant. External validation found our model to be generalizable. Our unique precision medicine approach can be applied to a wide range of patient populations and outcomes.
Insights
Researchers developed a novel machine learning method to predict multiple independent infection episodes (MIIEs) in trauma patients before they occur. This early detection of infection risk improves patient outcomes and reduces mortality.
Area of Science:
- Genomics
- Translational Medicine
- Computational Biology
Background:
- Severe trauma increases the risk of multiple independent infection episodes (MIIEs), significantly raising morbidity and mortality.
- Current methods detect infections only after they are established, limiting timely intervention.
- A proactive approach to identifying infection risk is crucial for improving patient care after trauma.
Purpose of the Study:
- To develop and validate a predictive model for identifying patients at high risk of MIIEs following severe trauma.
- To differentiate this predictive approach from existing reactive infection detection methods.
- To leverage machine learning and transcriptomic data for early risk assessment.
Main Methods:
- Machine learning algorithms (LASSO and Elastic Net) were applied to genome-wide transcriptome data from leukocytes of 128 adult blunt trauma patients.
- Data were collected within 48 hours of injury and at least 3 days before any clinical signs of infection.
- A 15-transcript and a 26-transcript multi-biomarker panel were constructed to predict MIIE risk.
Main Results:
- The developed models accurately predicted MIIE risk with high Area Under the Receiver Operating Characteristics Curve (AUROC) values (0.90 and 0.92).
- The predictive performance significantly surpassed existing clinical models.
- External validation confirmed the generalizability and robustness of the predictive model.
- Gene Ontology and network analyses identified relevant biological pathways associated with MIIE risk.
Conclusions:
- A novel, precision medicine approach using transcriptomic data and machine learning can predict MIIE risk in trauma patients before clinical manifestation.
- This method offers a significant advancement over current reactive infection detection strategies.
- The validated model shows potential for broad application across diverse patient populations to improve outcomes and reduce mortality.
