AI to detect fatigue – and how it affects performance under stress

DARPA funds $4.8 million biometrics project but findings will apply to general public, too.

August 22, 2022

2 Min Read

DARPA funds $4.8 million biometrics project but findings will apply to general public, too.

Researchers at Texas A&M and Arizona State universities received a $4.8 million grant to develop an AI platform to detect levels of fatigue by analyzing the human breath.

The project seeks to understand how physical and mental fatigue can affect performance, especially in stressful or high-stakes situations.

While funding is from the Defense Advanced Research Projects Agency’s (DARPA) Biological Technology Office, the model also will be used to build resilience for the general population.

“Fatigue is an important topic for the U.S. Department of Defense and many other sectors in our society. Yet, it is very challenging to quantify fatigue,” said project lead and Texas A&M professor Roozbeh Jafari.

The AI model works by evaluating breath volatile organic compounds (VOC) biomarkers. VOCs have been successfully used to detect asthma and bowel inflammation. The chemical compounds from all body systems can provide significant clues about a person’s overall state of health.

By evaluating and predicting fatigue, the scientists hope to develop new insights that could potentially be incorporated into wearables, which could help prevent nervous breakdowns and unearth new methods to build human resilience in difficult situations.

“Comprehensively examining the change in breath VOCs during the progression from rest to fatigue will provide valuable insights into the transitions in metabolic states,” said Steven Riechman, Texas A&M’s kinesiology professor who also is involved in the project.

The study will use wearable monitors and sensors to measure body temperature, heart rate, and other biometrics. More than 3,000 breath VOC samples will be generated. Breath samples at various stages of fatigue will be evaluated. The combination and volume of breath samples will contribute information to create models to predict fatigue.

Statistical modeling and pattern recognition will be applied to the VOCs to detect and categorize the types of fatigue. The project aims to develop physiological and breath biomarkers of fatigue not only for the military but also for the general public.

Get the newsletter
From automation advancements to policy announcements, stay ahead of the curve with the bi-weekly AI Business newsletter.