A special forces warrior in training endures a tremendous amount of physical challenges in hopes of obtaining acceptance into elite military programs, such as U.S. Army Rangers, Navy SEALs, or Marine Corps Raiders. Though these kinds of individuals may be top-tier in terms of physical fitness, their bodies all differ from each other and would benefit from personalized analysis and training.
In April 2020, the Georgia Tech Research Institute (GTRI) received a $1.6 million grant for Project SHARPI, which stands for “Supporting Human Adaptation for Readiness and Performance Improvements,” by the Office of Naval Research (ONR). The goal of the project is to advance the art of Machine Learning and Artificial Intelligence (ML/AI) in human performance and adaptation.
Researchers gathered physiological data from a variety of sources and military training sites, such as from U.S. Army Ranger trainees at Fort Benning. Using wearable sensors, researchers measured participants’ heart rates, skin temperatures, joint movements, gait, and other factors. Through this project, GTRI aims to use this data to create a “virtual coach” that could help inform military training.
“Most of the time, the way that [the military trains] people is purely based on previous knowledge and personal experience of the drill sergeant — what we call doctrinal guidance,” said Alessio Medda, a GTRI senior research engineer who is leading the project.
Medda emphasized the need for improving training, reducing injuries, and understanding the kinds of measurements that can be obtained from the wearable sensors.
“This is more than just your Fitbit-kind-of analysis of the number of steps that you’ve been taking, and motivating you towards 10,000 steps,” said co-principal investigator C.J. Hutto, a GTRI senior research scientist who is providing a human factor and behavioral data science perspective on the project, “— because these individuals are already very near the ceiling of physical fitness, and so the window of variability for improvement is very, very narrow.”
“In fact, the work we are doing here will produce a more complex, explainable, and nuanced perspective to aid the government through novel injury risk estimation and performance recommendations via new kinds of sensor combination analyses, coupled with the use of computational modeling,” added Scott Appling, a GTRI senior research scientist and co-principal investigator on the project.
Painting a Detailed Picture
Two sensors used to collect data for Project SHARPI are also found in almost every sports watch: the electrocardiogram (EKG or ECG), which measures heart rate, and the accelerometer, which measures motion.
“Those two sensors are very common, and doing the analysis and the software processing of those signals is fairly straightforward,” Hutto explained.
What is unique about Project SHARPI, however, is the way GTRI is combining data. Incorporating information from echocardiograms, accelerometers, measurements of skin temperature, and other types of sensors paints a more informative picture and provides physiological context.
One of the additional sensors is the electromyograph (EMG), which measures muscle activation.
“We can understand sort of at the micro muscle activation level whether or not [the trainees] are getting muscle fatigue during their exercise,” Hutto said. “And what’s interesting is that when we put that together with personalized data about the trainee, such as their age, gender, height, and weight along with the accelerometer data and the heart rate data, we can start to characterize much more complex kinds of physiological responses to the exercise stress.”
The EMG was just one of several sensors used to create a more detailed picture of physiology for military trainees. Others include innovative wearable acoustic sensors, which offer a very unique kind of biosignal for an even richer perspective that integrates data about joint health during intense training.
The Music of Motion
A professor at the Georgia Institute of Technology’s (Georgia Tech) School of Electrical and Computer Engineering has been studying the sound that joints make when in motion.
For several years, professor Omer Inan, director of the Inan Research Lab, has recorded the creaking, popping, and clicking on live participants and even on cadavers. He says working with cadavers allows researchers to induce injuries and then study the joint sounds to understand how an injury affects them.
More recently, however, Inan began studying the effect loading has on joint sound — like a soldier’s marching with a heavy weight, for example.
He’s also combining the joint sound data with cardiovascular data and providing his expertise to GTRI on how to accurately classify measurements and weed out anything that can corrupt it.
“For example, if you’re measuring heart-related parameters from a person who is still, it’s a lot easier than if they’re walking around, moving around, [or] walking uphill [or] downhill,” Inan said. “Those sorts of movements create artifacts that can corrupt the measurements.”
Inan emphasized the importance of looking at the bigger picture and combining his joint sound data with other measurements. Looking at the bigger picture is what can help soldiers train better and avoid injuries that could delay their progress.
“I think to truly capture performance — and potentially performance degradation due to injury — I think you need a holistic view of sensing,” Inan said.
Learning how to minimize injuries and maximize performance will not only benefit soldiers in training but also other physiologically advanced individuals, such as professional athletes.
People who train intensely face a higher risk of injury than those who only exercise to stay fit. Consequently, they require a more personalized and microscopic approach to their training due to their narrow window for potential improvement.
Broader applications of the “virtual coach” may be explored in the future, but right now, GTRI is committed to improving the training techniques of our military personnel so they can complete their jobs with excellence.
Article credit: Kaitlyn Lewis