Undernutrition prevention for disabled and elderly people in smart home with Bayesian networks and RFID sensors

Undernutrition prevention or detection for disabled or elderly people must be performed rapidly to avoid irremediable consequences. In this paper a classification of uncertainties centered on a meal notion is first proposed. Two of these uncertainties are developed in a smart home and homecare context. Meal preparation probability is evaluated by a simulation based on Nauve Bayesian Networks. To determine if a person is at risk of malnutrition or undernutrition, and to supervise prepared meal quality and quantity in terms of nutrients, the use of RFID tags is discussed, bringing many open issues for which additional sensors are proposed. This research work was initiated in a collaborative project called CaptHom.