MedDietAgent: An AI-based Mobile App for Harmonizing Individuals' Dietary Choices with the Mediterranean Diet Pattern.

Abstract

Recently, there has been an increasing interest in applying technological advances to offer specific dietary recommendations in the field of nutrition and health. Dietary recommendation systems are advanced tools designed to assist individuals in making well-informed and health-conscious decisions on their food choices, taking into account their personal needs, preferences, and health targets or habits. In this study, we present an AI-based mobile app for harmonizing individuals' dietary choices with the pattern of the Mediterranean diet. A combination of computer vision, natural language processing, machine learning, and reinforcement techniques are used to record the nutritional information via images or speech and to generate dynamic recommendations tailored to the user's performance across key nutritional areas, encompassing calories, combined fats, proteins, carbohydrates, sugars, dietary fibers, sodium intake, fruits, vegetables, and dairy products. The image-based dietary assessment subsystem achieves a mean absolute percentage error of 3.73%, while the reinforcement learning subsystem achieves a 96% average reward. Then, a well-designed approach was taken to develop the MedDietAgent mobile app, using cutting-edge technologies and applying a simplistic approach. One of the key aspects of MedDietAgent is its ability to offer dynamic recommendations by monitoring the user's environment.