A robust optimization approach to diet problem with overall glycemic load as objective function

The main purpose of the paper is to introduce a mixed-integer programming model for the diet problem with glycemic load (GL) values of foods as objective function parameters. It is assumed that the glycemic load values are subject to uncertainty. The diet problem with minimum cost function is well-known in the literature. However, the diet problem with minimum total daily GL values of foods that satisfies the daily nutritional and serving size requirements has not been proposed. Robust optimization approach is used to account for uncertainty in the GL values of foods. The decision maker is flexible to tune the degree of uncertainty rather than assuming a worst-case scenario. An experimental analysis with a total of 177 foods is performed based on the nutritional and serving size requirements and the basic food groups recommended by the U.S. Department of Health and Human Services & U.S. Department of Agriculture (USDA). The results of the experimental analysis with different scenarios give different solutions for different degrees of uncertainty. However, some foods are frequently found to be in the optimum solutions. These foods are in good agreement with the literature advising them as a part of a daily diet for attaining low level of blood glucose levels. Although we believe that the proposed diet problem with minimum total GL has contributions for satisfying the daily nutritional and serving size requirements with a minimum level of effect on blood glucose levels, it has several limitations. It is a basic diet problem, and assumes that the overall GL is a linear combination of number of serving sizes with the GL values of foods. It also does not consider any other factors such as several combinations of foods and their varying effects on blood glucose levels. These factors should be considered for the next research.