LINEAR PROGRAMMING TECHNIQUES FOR THE CONSTRUCTION OF PALATABLE HUMAN DIETS

Abstract

From the earliest days, electronic computers have been applied to calculations on human diets, and there are now software packages which virtually eliminate the computational drudgery in obtaining a nutritional analysis of a diet. However, dietitians have long worked with the reverse problem; namely, advising patient what they should eat and in what quantities, to meet the nutrient intake levels recommended for either general health or disease-related reasons. Many textbooks on linear programming mention the diet problem; that is the divising of a diet which meets nutrient targets at minimum cost. This is seen as an illustration of mathematical techniques rather than a contribution to human dietetics, although the approach is widely used in formulating animal feedstuffs. The pioneer of work on human dietary optimization was George Stigler who showed how to feed a grown man for one year at a total cost. This 'optimal' diet was, needless to, say very restricted, boring and unpalatable. Subsequent work has incorporated constraints, additional to the strict requirements of nutrition, to increase the variety and acceptability of the diet. A benchmark is provided by a study in the US Mid-West which constructed nutritious and economical diets with a range and variety similar to actual eating patterns.