Mass casualty incidents (MCIs) impose extreme
cognitive and perceptual demands on first responders. Traditional START triage requires personnel to recall a sequence of
checks while managing multiple casualties, leading to error rates
between 20 % and 55 % and long assessment times [1]. Smart
glasses with augmented reality (AR) workflows can improve
triage accuracy and reduce time-to-triage by more than a
factor of two [1], but previous deployments have suffered from
usability issues and excessive cognitive load [2]. We present
Triage-AR, a human-in-the-loop casualty visualization system that
delivers priority-aware overlays through AR headsets. Triage-AR
computes dynamic priorities for each casualty using physiological
and situational data, and uses colour, haptic and audio cues
to emphasise the most critical patients while suppressing less
urgent information. Our user study with 24 participants in
simulated arena scenarios shows that priority-aware overlays
reduce median time-to-triage by 27 % relative to a baseline AR
overlay, with significantly fewer missed casualties and lower
NASA-TLX workload scores. The results support the claim that
priority-aware overlays can achieve at least a 25 % reduction in
time-to-triage while preserving mission utility. We open-source
our software and datasets to encourage reproducibility.