Poster Topical Area: Methods and Protocols
Location: Hall D
Poster Board Number: 681
Objectives: Choosing an optimal combination of foods included in a daily eating plan (DEP) can be confusing, difficult, and time consuming, especially when more than a few chosen characteristics of foods are considered, e.g. calories, protein, fats, and sodium. To achieve this aim, a method was developed using genetic algorithms (GAs), a metaheuristic computational technique based on evolution, to find near-optimal DEP solutions that can include several nutritional goals.
Methods: A method is described to use GAs to quickly create a custom DEP, drawn exclusively from a "pantry" of foods chosen by the user on the basis of their palatability, availability, and affordability. Any foods for which nutritional data exist can be included in the user's "pantry" and, therefore, in the DEP. While determining the globally optimal DEP might require years of computational time due to the huge number of combinations of foods which must be considered, GAs can pare the task efficiently and thereby generate a near-optimal solution in only seconds. This protocol is particularly useful when seeking to optimize a DEP that is constrained by several simultaneous goals and/or dietary restrictions. Furthermore, this method may be used to create a de novo eating plan each day or to determine a near-optimal DEP that includes foods that have already been eaten that day. The nutritional goals and foods in the "pantry" are easily configurable, all food and preference data can be retained from session to session, and a different solution can be generated in seconds if needed.
Conclusion: A method is disclosed for rapidly and easily determining a near-optimal DEP which requires consideration of multifactorial nutritional goals through the use of GAs, all of which can be customized by a professional or lay user.
School of Engineering and Applied Sciences, Harvard University