Tracking wrist motion to detect and measure the eating intake of free-living humans

This dissertation is motivated by the growing prevalence of obesity, a health problem currently affecting over 500 million people worldwide. It is composed of two studies. In the first study, a new method is developed to detect how many bites a person takes during a meal in real time. A pattern has been found that the wrist of a person undergoes a characteristic roll motion as food is picked up and placed into the mouth. This motion can be tracked by a gyroscope sensor placed on the wrist. This work could be used in many weight loss and obesity treatment applications, including monitoring intake, slowing eating rate, and providing a cue for mindful eating. In the second study, a new method is developed to automatically distinguish eating activity from other activities in natural daily living. Accelerometers are used to detect the typical burst activity at the beginning and the end of each eating activity and gyroscope roll motion features are used during hypothesized detections to differentiate eating activities from other activities. This work has many potential applications. It could be used by individuals for self-monitoring for weight loss and weight maintenance. It could be combined with a food diary, 24-hour recall or food frequency questionnaire to improve compliance and accuracy in measuring consumption. The two methods could potentially be combined to automatically count bites of intake all day. The methods could also be used by clinical practitioners to monitor the eating patterns of patients (for example during diabetes treatment), or by researchers in epidemiological and genetic studies (for example in studies of the physical activity or eating habits of specific demographics).