Linear programming techniques for the construction of palatable human diets

Mathematical modelling of the relationship between acceptable diets and satisfactory nutrition is discussed. The classical minimum-cost diet problem is unsuited to most purposes in human dietetics, since the objective of least monetary cost leads to diets which are unpalatable, both generally and to individuals. In this paper a different approach, using an objective function based on the food preferences of individuals, is described. A computational method, based on this analysis, for constructing individually acceptable diets by means of linear programming is described. Careful use of post-optimal analysis provides invaluable insight to individual patients, and their medical advisors, on the changes to their eating habits which will most readily enable them to meet nutrient intake targets. An example, using data appropriate to a patient with chronic renal failure, demonstrates the usefulness to nutritionists and dietitians of this approach to dietary recommendations.