Sistema de información nutricional y recomendación de sustituciones saludables de productos procesados para la prevención de DMT2 en Cuenca

Abstract

Type 2 Diabetes Mellitus (T2DM) is one of the biggest threats to Ecuador's health. The consumption of processed foods has been linked to a higher risk of acquiring T2DM. Therefore, efforts to discourage the consumption of these products are necessary. This paper proposes FoodSub, a mobile application to recommend food substitutes to replace processed foods using the NOVA Classification. The proposed approach uses nutrient-based food clustering to identify substitute pairs between processed and unprocessed foods. The recommendation of food substitutes is supported and personalized using a knowledge graph that contains foods, dietary guidelines, and user information. In addition, a chatbot is implemented to answer simple questions about food nutritional values, comparisons, ingredients, and general information. This chatbot is developed using a Large Language Model (LLM) to query the implemented knowledge base. The mobile application and the chatbot are evaluated in terms of usability; both perform well according to the System Usability Scale (SUS) and the Chatbot Usability Questionnaire (CUQ), respectively, but there is some scope for improvement. Additionally, the substitute recommendations performance is evaluated through expert verification. The results indicate that the recommendations perform well in cases where the differences between the original food and the substitute are not relevant in terms of food transformation processes, flavor, context, or meal time. Future work lines are provided, considering the enhancement of the chatbot and the improvement of substitute recommendations when the mentioned issues are relevant.