Field AI tool triages victims of gun violence

Emergency surgery specialists have developed a deep neural network for triaging patients presenting with gunshot wounds.

In a tryout trial, the technique quickly and accurately predicted shock as well as the need for early massive transfusion and major surgery.

The research was conducted at Massachusetts General Hospital and is running in the June edition of the Journal of Trauma and Acute Care Surgery.

Senior author Noelle Saillant, MD, and colleagues trained their field-capable AI tool on data from close to 30,000 patients with gunshot wounds to the trunk or appendage junctions thereof.

Along with wound locations and apparent extents, the training data included vital signs and other basic information that’s usually readily available in the field.

In testing, the tool had a high degree of accuracy predicting shock, transfusion need and surgery urgency, Saillant and co-authors report.

“The presented model is an important proof-of-concept,” they write. “Future iterations will employ an expansion of databases to refine and validate the model, further adding to its potential to improve triage in the field, both in civilian and military settings.”

Study posted here (behind paywall).

Around the web

The vast majority of healthcare executives say digital transformation across their organizations is accelerating, according to a recent survey of 399 executives from six countries.

Reperforming lateral knee radiographs is common practice but consumes unnecessary resources and exposes patients to added radiation, experts explained in Radiography.

An estimated 100,000 people in the U.S. live with sickle cell disease, which disproportionately affects African Americans.

These microdiamonds can be detected simultaneously via optical and radiofrequency imaging methods, opening up a number of new possibilities.