Paulo Lopez-Meyer, Stephanie Schuckers, Oleksandr Makeyev, Edward L. Melanson, Michael Neuman and Edward Sazonov, Annals of Biomedical Engineering, 2010, Volume 38, Number 8, 2766-2774, DOI: 10.1007/s10439-010-0019-1



Obesity is a growing problem in the world. Study of energy intake and ingestive behavior in free living conditions most often relies on self-reporting methods that can be highly inaccurate. Methods of Monitoring of Ingestive Behavior (MIB) rely on objective measures derived from chewing and swallowing sequences and thus can be used for unbiased study of food intake in free living conditions. Our previous work demonstrated accurate detection of food intake in simple models relying on observation of both chewing and swallowing. This paper investigates methods that achieve comparable accuracy by using only the time series of swallowing events and thus eliminating the need for the chewing sensor. The classification is performed for each individual swallow in the sequence rather than previously used time slices (epochs) and thus will lead to higher accuracy in mass prediction models relying on counts of swallows. Performance of a group model based on supervised (Support Vector Machines) method is compared to performance of an individual model based on unsupervised (K-means clustering) method with results indicating better performance of the latter. Overall, the results demonstrate that highly accurate detection of food intake is possible by an unsupervised system that does not need training and relies on the information provided by the swallowing alone.



Full text at SpringerLink