TheMachine Press

A daily newspaper for the age of artificial intelligence.

Morning editionPermanent story

research

Trauma Triage Improves When the Model Advises, Not Decides

A University at Buffalo study found clinicians made more accurate calls when an LLM recommendation accompanied noisy emergency communications.

Published Updated Story ID: mp-2026-07-12-015
Read the complete editionStory JSON

Summary

A University at Buffalo study found clinicians made more accurate calls when an LLM recommendation accompanied noisy emergency communications.

Language models could match or slightly exceed human interpretation of prehospital trauma calls, while clinicians performed best when they saw a model recommendation alongside their own judgment. The result supports decision support, not autonomous triage, and needs prospective clinical validation.

Why it matters

A University at Buffalo study found clinicians made more accurate calls when an LLM recommendation accompanied noisy emergency communications.

Limits and context

  • The result supports decision support, not autonomous triage, and needs prospective clinical validation.

Key claims

  1. A University at Buffalo study found clinicians made more accurate calls when an LLM recommendation accompanied noisy emergency communications.

    Qualification: The result supports decision support, not autonomous triage, and needs prospective clinical validation.

    Evidence: source-2026-07-12-015

Sources

  1. University at Buffalo: Trauma triage and LLM assistancewww.buffalo.edu · official announcement

Corrections

No corrections have been recorded for this story.