Abstract
The complexity surrounding food selection is attributable to the variability in foods, restaurants, and diners. The diversity of foods, where each dish may have a unique recipe across different restaurants, needs to be accounted for in personalized nutrition. However, personalized food selection poses a combinatorial challenge in selecting the most suitable food at a specific restaurant. The key question is how a diner visiting a particular restaurant can be assisted in selecting optimal foods and beverages based on factors such as sex, age, height, weight, and history of non-communicable diseases (NCDs). In this study, a genetic algorithm (GA) is used to develop a system that can address this issue in the context of Indonesian restaurants. In this system, a database with data on registered diners and foods is maintained. Foods comprise staple foods, side dishes, vegetables, and beverages, each containing its energy and nutrient content for a given restaurant. The nutritional adequacy of a single meal is determined by comparing the energy and nutrient content of the menu with the diner's nutritional needs. The novelty of the proposed system lies in combining scientific nutritional data with individual diner profiles for the selection of the best meal for a diner. This system differs from the existing food recommender applications in Indonesia, which typically do not consider specific diners, personalized nutrition, and NCD history. The proposed system is the first developed application prototype for Indonesian restaurants to overcome the inefficiency of the existing applications. In this study, the structure and chromosome content of the food, its corresponding energy and nutrient contents, and GA operators such as crossover, mutation, and tournament selection for determining the best meal using the defined fitness functions are discussed. The proposed system has been tested at Karimata Restaurant and proved to be highly suitable for the ultimate goal of meal selection for individual diners with different needs, and it can be replicated at other restaurants. Furthermore, user-centered evaluation has revealed that the system (a) increases nutritional understanding and health awareness; (b) is easy to use with comprehensive functions; and (c) promotes user satisfaction with personalized recommendations.