A computerized system for instruction in food selection practice

This thesis has developed arprototypical system which provides information on dietary practices for those individuals interested in applying nutritional principles to their eating habits. The system has the potential to provide information which both accurately reflects nutritional guidelines and facilitates adoption of recommendations, by providing a self-explanatory statement of foods to consume and by limiting suggested changes in present food pattern. The prototypical computerized system developed has two major functions: (i), diet-assessment to appraise the acceptability of individual's dietary practices; and (ii), diet-planning to recommend modifications in the diets of those individuals not meeting specified limits. The focus of the system is a constrained-optimization algorithm that generates a revised food plan which both satisfies nutrient constraints, and minimizes the deviation of food items rand item groups from the original amount consumed by the client. Testing has been restricted to a descriptive evaluation of some of the algorithm's characteristics -- specifically, the design assumptions which define the acceptability of deviating from an original inventory, and the revised diets developed when these assumptions are modified. The results illustrate that altering these design assumptions produces marked variations in the revised diets with respect to observed parameters. Further modifications in the algorithm have been suggested. The explorative evaluation provides a foundation for more systematic evaluation of the validity of the algorithm. Recommendations for facilitating the further development and testing of the system are outlined. This thesis has shown that mathematical modeling provides an effective means of collating the vast amount of data required to develop cogent dietary recommendations which are nutritionally accurate, straightforward, and acceptable to the client.Land and Food Systems, Faculty ofGraduat