Balancing meals using fuzzy arithmetic and heuristic search algorithms

This paper aims at showing how well-known ideas in the fields of fuzzy arithmetic and heuristic search have been combined in an educational software in nutrition in order to provide not only a better mathematical modeling, but also significant functional improvements for end-users, comparing to other nutrition programs. This software, called Nutri-Expert, helps patients to improve their nutritional habits, by analyzing in detail their food intakes, and by suggesting changes that result in well-balanced meals. Fuzzy arithmetic is used to model the input and database data, and for all computations. A fuzzy pattern matching is performed between total amounts of nutrients and different norm patterns, and the results are displayed using a galvanometer metaphor. A heuristic search algorithm is used to find out minimal sets of pertinent actions to perform on a meal in order to make it well balanced. The search is guided by an evaluation function based on fuzzy pattern matching indexes. The different versions of the algorithm have been benchmarked against a test database of real meals. Finally, the medical efficacy of Nutri-Expert and its acceptance by end-users have been demonstrated in several medical studies, the main results of which are presented.