The researchers say the new study is the first to investigate aerobic system dynamics with machine learning and unsupervised wearable sensor use.
When most people think of wearables, they typically think of wristband monitors and smartwatches. But there’s also things like “smart shirts,” actual garments that contain sensors for heart rate, breathing, and motion.
Researchers at the University of Waterloo in Ontario, Canada, recently used such shirts to see if they could develop an algorithm to detect early signs of future chronic diseases. They first studied 13 healthy men in their 20’s in a laboratory-based fitness program, creating metric benchmarks. The men then wore the shirts in their daily lives for 4 unsupervised days.
They found that the fitness characteristics measured during daily life correlated closely with those set during the laboratory sessions, and by combining all of those characteristics based they were able to create what researcher Richard Hughson called a “meaningful single number to track fitness.”
Early detection of subtle aerobic system impairments could help tip off healthcare providers to negative changes in a patient’s health. It could also allow patients with conditions like type 2 diabetes and chronic obstructive pulmonary disease to constantly monitor their own fitness and disease state.
Alexander Wong, an artificial intelligence and engineering expert at Waterloo, worked with Hughson, a kinesiology professor at the Schlegel-University of Waterloo Research Institute for Aging, and Thomas Beltrame, a computing expert who has since begun working at the University of Campinas in Brazil. The collaboration was key to developing the algorithm.
"This multi-disciplinary research is a great example of how artificial intelligence can be a potential game-changer for healthcare by turning data into predictive knowledge to help healthcare professionals better understand an individual's health," Wong said. "It can have a significant impact on improving quality of life and well-being."
The work is still early—it refers to itself as “speculative”—but future studies will focus on whether the smart shirts and algorithms maintain their predictive prowess for people of other ages and genders, and those suffering from chronic diseases.
The report was published this week in the American Physiological Society’s Journal of Applied Physiology. The authors claim it is the first study “to use wearable sensors in unsupervised activities of daily living in combination with novel machine learning algorithms to investigate the aerobic system dynamics.”