URI professors monitor eating with tracking device
Have you ever wondered how your eating habits align with other daily activities that are tracked — calories, steps, blood pressure?
University of Rhode Island professors Kathleen Melanson and Theodore Walls are working with University of Texas at Austin professor Edison Thomaz to develop devices that will track eating behavior through machine learning. Their study is funded by a National Institutes of Health grant and is a collaboration of their backgrounds in nutrition, behavioral statistics, and engineering.
Q: Where did the idea for the study come from?
Melanson: I was working on understanding the nuances of eating behavior that affect energy intake in humans. Laboratory studies are very precise because we can measure very precisely in the lab, but there’s nothing to do it that precisely in the real world. That was limiting our ability to really understand what’s going on with eating behavior. We wanted to take the tools outside of the lab.
Q: What does the study entail for participants?
Walls: It’s actually four substudies over four years. They start out in the lab and they move to the real world, increasing in complexity over time. The studies are experiments in the lab to make sure we know where the device goes and how well it can work without difficulty for people in rather controlled situations. Then they increase in complexity to things like cafeterias or meals together with other people and different foods.
Q: What types of devices will participants receive?
Walls: The devices themselves are under development as part of the funding that we received. A name for this kind of device is called multi-modal sensing. We’re trying to figure out which sensors will work together best in a very small area to sense the movements of the jaw. The target region that we propose is called the condyle bone area [the round prominence at the end of a bone]. We believe that the area above the condyle bone will move in a systematic way in relation to opening and closing of the jaw. We’re hoping, actually, that we can get into very detailed information about the strength of plates.
Q: And what about the smartwatch?
Melanson: The wrist motion for placing food in the mouth is very specific, so the device can just differentiate that more from other similar movements. When it’s coupled with the chewing sensor, the accuracies improve, because it’s coupling the food placement in the mouth with the chewing. It also helps us learn better about the nuances of eating behavior that matter in the development of satiety and what we know is associated with people’s energy intake.
Q: Why is this study important to your field of work?
Melanson: I’ve been working in the field of nutrition for decades, and specifically looking at eating behaviors for promoting health and well-being of individuals. Much of the data in this area is based on laboratory research, in which I’ve been conducting for decades. We have tried to take it outside the lab, but it requires a participant to record everything in full detail, and it’s very time-consuming. So we needed something that was like a Fitbit, something that is automated. You just wear it. You don’t have to do the work yourself. This is becoming a realization that, yes, this can happen for eating behaviors in addition to other health metrics that are tracked like physical activity.
Q: What benefits can come from the study?
Melanson: We can start designing programs for people who want to improve their eating patterns — so people who struggle with loss of control, rapid eating, or eating without awareness. The awareness factor of this can be beneficial for people who struggle with late nights and those type of things. I envision a tool that can be used to help people who are seeking to improve these behaviors, but also their overall dietary intake.
Q: What’s your main goal coming out of the study?
Melanson: We finally can take this research out of the lab and deploy it in people who are seeking to improve their health behaviors, and help give them a tool that they can use. There’s no automated tool for eating behaviors yet.
Q: How can people sign up to be participants in the study?
Walls: Anyone interested in being a participant in the study can contact the researchers at dibs@uri.edu.