Development of a Fuzzy Expert System for a Nutritional Guidance Application

Abstract

The importance of nutritional guidance grows as nutritional problems, such as obesity and type-2 diabetes, are becoming more common. Nutritional guidance is carried out by mapping the nutritional state of an individual using a food diary and by comparing the nutrition intake levels to the recommended reference values. Typically, the expert knowledge of a nutritionist is required to balance the diet. This paper presents a fuzzy expert system for a nutritional guidance application. Expert knowledge acquisition is carried out using variable tabulation creating the basis for the rulebase of the fuzzy system. The recommended nutrition intake values are used to generate membership functions for fuzzification. The development of a Mamdani-type fuzzy system is illustrated using a hierarchical structure. The first level system refers to groups of similar foods and the second level model defines the added and reduced foodstuffs. The validation of the fuzzy model was carried out in three phases: first two types of sensitivity analysis were performed, and then the output was analysed with expert knowledge.. The results from the validation schemes are promising.