Applied Research Associates, Inc. (ARA) is leveraging its technical expertise to improve battlefield medicine by tackling some challenges related to assessing thermally-injured causalities.
The Burn Medical Assistant (BurnMan) project is a collaboration between ARA and the U.S. Army Institute for Surgical Research (USAISR). The objective of BurnMan is to develop machine learning algorithms to automate the process of analyzing and classifying burn wounds on the battlefield.
It is not unusual for an inexperienced medic to be 50 percent or more in error when calculating the total body surface area of a burn, and even experienced medical doctors can vary by 10 to 20 percent on the same patient.
ARA is using multi-spectral image data and advanced machine learning algorithms to automate the process of assessing burns and to generate a burn wound map, called an enhanced Lund and Browder diagram, that will significantly reduce the potential for error when a medic is assessing a burn casualty.
The BurnMan prototype will integrate advanced imaging, burn wound mapping technology, and machine learning to provide novice burn care providers with an estimate of burn wound size to within five percent of an expert’s estimate. This information can then be used to provide support and information to develop an appropriate care plan.
“While still early in the research, this technology could significantly improve the ability of medics to assess and document burn casualties,” said Principal Investigator Chris Argenta, principal software engineer and expert in artificial intelligence (AI) and machine learning (ML). Dr. Argenta leads ARA’s Decision Systems Group, which is heavily involved in advanced AI/ML research across Department of Defense, intelligence community, and medical domains.
This research was recently presented by Project Manager Greg Rule, a principal engineer with ARA, at the Military Health Systems Research Symposium in Orlando, FL, where it received honorable mention as one of the top 15 posters out of over 1,000 posters presented at the event.