Prehospital triage of trauma patients using the Random Forest computer algorithm.

Area of Science:

• Emergency Medicine
• Trauma Surgery
• Health Informatics

Background:

• Overtriage of trauma patients leads to wasted resources and delayed care for critically injured individuals.
• Accurate patient triage is essential for efficient trauma system operation.

Purpose of the Study:

• To validate a Random Forest computer model (RFM) for improved trauma patient triage to Level 1 trauma centers.
• To assess the RFM's ability to reduce overtriage rates.

Main Methods:

• Adult trauma patients with "medium activation" criteria were analyzed.
• The RFM incorporated demographics, injury mechanism, prehospital care, and vital signs.
• The RFM's triage decisions were compared against the institution's historical triage rates.

Main Results:

• The RFM reduced overtriage from 66% to 42% (P < 0.001).
• The model correctly identified 88% of patients requiring Level 1 trauma center care.
• An undertriage rate of 8.3% was observed with the RFM.

Conclusions:

• Computer modeling, such as the RFM, shows potential for guiding trauma triage decisions.
• Implementing such models could lead to more accurate triage and efficient trauma system utilization.
• Prospective validation is necessary to confirm these findings.