DIET PLANNING FOR HUMANS USING MIXED-INTEGER LINEAR-PROGRAMMING

Human diet planning is generally carried out by selecting the food ite ms or groups of food items to be used in the diet and then calculating the composition. If nutrient quantities do not reach the desired nutr itional requirements, foods are exchanged or quantities altered and th e composition recalculated. Iterations are repeated until a suitable d iet is obtained. This procedure is cumbersome and slow and often leads to compromises in composition of the final diets. A computerized mode l, planning diets for humans at minimum cost while supplying all nutri tional requirements, maintaining nutrient relationships and preserving eating practices is presented. This is based on a mixed-integer linea r-programming algorithm. Linear equations were prepared for each nutri tional requirement. To produce linear equations for relationships betw een nutrients, linear transformations were performed. Logical definiti ons for interactions such as the frequency of use of foods, relationsh ips between exchange groups and the energy content of different meals were defined, and linear equations for these associations were written . Food items generally eaten in whole units were defined as integers. The use of this program is demonstrated for planning diets using a lar ge selection of basic foods and for clinical situations where nutritio nal intervention is desirable. The system presented begins from a defi nition of the nutritional requirements and then plans the foods accord ingly, and at minimum cost. This provides an accurate, efficient and v ersatile method of diet formulation.