CASE-BASED REASONING FOR NUTRITION CONSULTING

Case-Based Reasoning (CBR) is an Artificial Intelligence (AI) paradigm, in which new problems are solved by storing, retrieving, and adapting solutions to previously encountered problems. The expert task of prescribing a nutritional program as therapy for a given disorder, individualized for each patient, reuses past experiences to generate the best possible solution. In many AI systems including Expert Systems, Semantic Network, Genetic Algorithm, Neural Network, although every diet prescription is an experience, they do not make use of past knowledge. In CBR, nutritional experiences are manipulated within the system and enables past knowledge to support nutritional practice. The phases of the project include prototyping, acquiring cases, building prototypical memory, indexing, elicitation of adapted knowledge and programming the final version. A sample CBR system was built and was composed of 600 real cases (393 women, 207 men, 20 to 65 years, healthy and chronic diseases e. g. heart disease, hypertension, diabetes, obesity). The prototypical memory was built for 15 prototypes. Two important aspects of the CBR system are knowledge acquisition and adaptation of prototypes. These problems were solved through use of a prototypical memory (representing different nutritional disorders) and by incorporating past solutions (nutritional plans) for a specific disorder into the updated prototypical memory. The current implementation produces diets based on past experiences, following the steps described. The model is being used clinically to evaluate its performance.