AI-Powered Bite Counter Could Help Combat Childhood Obesity
A new artificial intelligence (AI) model developed by Penn State researchers shows promise in accurately measuring how quickly children eat, a key factor linked to obesity risk. The system, dubbed ByteTrack, represents a significant step forward in overcoming the challenges of studying eating behaviors, which traditionally relied on painstaking manual analysis of video footage.
Researchers have long known that faster eating speeds are associated with a higher risk of developing obesity. However, quantifying this connection has been difficult due to the labor-intensive nature of tracking a child’s bite rate – the frequency at which a child takes bites during a meal or snack. Traditionally, this required researchers to meticulously watch videos and manually record each bite.
To address this obstacle, a collaborative team from the Penn State Departments of Nutritional Sciences and Human Development and Family Studies created an AI model capable of automatically measuring bite rate. The results of a recent pilot study, published in Frontiers in Nutrition, indicate the system currently achieves approximately 70% of the accuracy of human bite counters.
“When we eat quickly, we don’t give our digestive track time to sense the calories,” explained Kathleen Keller, professor and Helen A. Guthrie Chair of nutritional sciences at Penn State and a co-author of the study. “The faster you eat, the faster it goes through your stomach, and the body cannot release hormones in time to let you know you are full. Later, you may feel like you have overeaten, but when this behavior repeats, faster eaters are at greater risk for developing obesity.”
Previous research from Keller’s laboratory group has demonstrated that a faster bite rate, particularly when combined with larger bite sizes, correlates with higher obesity rates among children. Other studies have also highlighted larger bite size as a potential risk factor for choking.
“Bite rate is often the target behavior for interventions aimed at slowing eating rate,” said Alaina Pearce, research data management librarian at Penn State and a co-author of the research. “This is because bite rate is a stable characteristic of children’s eating style that can be targeted to reduce their eating rate, intake and ultimately risk for obesity.”
The development of ByteTrack was spearheaded by Yashaswini Bhat, a doctoral candidate in nutritional sciences, who collaborated with Timothy Brick, associate professor of human development and family studies at Penn State. Bhat, with an interest in AI and data science, sought to create a more efficient method for analyzing children’s eating habits.
The researchers trained the AI model using 1,440 minutes of video footage from the Food and Brain Study, a National Institute of Diabetes and Digestive and Kidney Diseases-funded project examining the neural mechanisms influencing overeating in children. The footage featured 94 children, aged seven to nine, consuming meals on separate occasions.
Initial testing showed the system excelled at identifying children’s faces – achieving 97% accuracy – but lagged slightly in accurately counting every bite. The model struggled with instances where a child’s face was partially obscured or when they engaged in behaviors like chewing on utensils.
“The system was less accurate when a child’s face was not in full view of the camera or when a child chewed on their spoon or played with their food,” Bhat noted. “As one might imagine, this type of behavior is much more common among children than it is with adults.”
Despite these limitations, researchers are optimistic about ByteTrack’s potential. With further refinement and training, they believe the system can become a valuable tool for researchers, and eventually, for parents and healthcare professionals.
“The eventual goal is to develop a robust system that can function in the real world,” Bhat said. “One day, we might be able to offer a smartphone app that warns children when they need to slow their eating so they can develop healthy habits that last a lifetime.”
This research was funded by the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of General Medical Sciences, the Penn State Institute for Computational and Data Sciences, and the Penn State Clinical and Translational Science Institute. At Penn State, researchers are dedicated to solving critical challenges impacting health, safety, and quality of life globally, and continued federal support is vital to sustaining this innovation. Those concerned about the impact of potential federal funding cuts can learn more at Research or Regress.
