Multi-Constraint Particle Swarm Optimization Algorithm in Diet Recommendation for Indian Elderly Persons

A meal recommendation plan should include varieties of food from different categories, and a proper balance of all the nutrients in a required amount. This is a combinatorial optimization problem and is NP-hard. To address NP-hard optimization problems, metaheuristic algorithms having good exploration and exploitation properties are used. This paper proposes a Multi Constraint Particle Swarm Optimization algorithm for meal recommendations for Indian elderly persons. The method adapted easily works with traditional PSO without extra computation cost. It evaluates fitness function based on various nutrient constraints from different categories. Moreover, to improve the particles' searching efficiency and to improve the quality of the solution, initialization of position and velocity of particles are done randomly from a customized set that includes personalised data obtained during registration.Experiments at an initial level were also conducted, showing positive results.