FOOD INTAKE MONITORING FOR PERSONS WITH ALZHEIMER’S USING SKIN DETECTION METHODS

摘要

An estimated 5.7 million individuals have Alzheimer’s disease and related dementias in the U.S., a figure expected to rise to 14 million by 2050. One of the major challenges caring for this population is to maintain adequate nutritional intake. Finding effective ways to support feeding activities is essential because malnutrition is linked to premature disability and diminished quality of life. Persons with Alzheimer’s are considered at high risk for malnutrition resulting from decreased appetite, attention deficits and an increasing inability to recognize food. We designed a computer-tablet device to monitor food intake that also has a prompting feature to encourage eating. The algorithm can distinquish eating from non-eating gestures within the upper body region using bounded boxing and a skin detection system. Our testing indiciated a high level of success. For this project 33 videos that included 920 eating events resulted in an accuracy rate of 90.65% in the lab setting. Beta testing involving two participants with Alzheimer’s disease indicated that both lighting and body positioning are important for accurate detection of eating. The participants were not disturbed by the device which was placed within the participant’s arm’s reach. Testing of the device is underway assessing the efficacy of a prompting video with the device, encouraging the individual to continue eating. Both informal and informal care providers are heavily involved in feeding individuls with cognitive loss. Developing approaches for promoting independent feeding as long as possible are essential to support this rapidly growing population of persons with Alzheimer’s disease.