Gaze Track

The Effect of Triage Training on Triage Classification Using Gaze Tracking

Simple Triage and Rapid Treatment (START) protocols are used during mass casualty incidents to sort victims into groups prioritized for medical treatment. The standard triage classifications are:

  1. Green/Minor: minor injuries;
  2. Yellow/Delayed: medical attention may be delayed for up to 1 hour;
  3. Red/Immediate: immediate (within 60 minutes) medical attention required;
  4. Black/Expectant: unlikely to survive or deceased.

This study evaluated how training in START triage protocols might affect how subjects interact with and make triage decisions, based on images of disaster victims. We hypothesized that subjects who received training (the experimental group) would alter the duration of fixation on salient features versus non-salient features in the images, when compared to subjects who received no training (the control group) in the START protocols. It was anticipated that the experimental group would display increased triage accuracy when compared to the control group.

Two groups of 12 subjects were used: the experimental group received triage training on ABC’s (Airway; Breathing rate; Circulation), the control group received unrelated, yet comparable, training on patient transportation. BeGaze software and SMI RED camera (120-Hz) tracked subjects’ gaze upon distinguished Areas of Interest (AOI) within each image. Text boxes were embedded on each image for ABCs and assigned an AOI with area outside these indicated as the White Space AOI. Post-training, the experimental group fixated upon educationally salient features significantly earlier than to the control group. The experimental group also spent significantly less time fixating on the white space AOI and more time on ABC AOIs. The control group demonstrated a statistically significant increase in first fixation duration on the white space AOI. Both groups showed improved triage accuracy during second viewings of the images, however due to a small sample size, there was no statistically significant finding.