A decision tree–based classifier to provide nutritional plans recommendations

• Autores: Aguilar-Loja, Omar, Dioses-Ojeda, Luis, Armas-Aguirre, Jimmy, Gonzalez, Paola A.
• Formato: artículo
• Fecha de Publicación: 2022
• Institución: Universidad Peruana de Ciencias Aplicadas
• Repositorio: UPC-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorioacademico.upc.edu.pe:10757/660550
• Enlace del recurso: http://hdl.handle.net/10757/660550
• Nivel de acceso: acceso embargado
• Materia: Decision tree classifier; Machine learning; Nutrition; Recommender system; Scikit-Learn

The use of machine learning algorithms in the field of nutritional health is a topic that has been developed in recent years for the early diagnosis of diseases or the recommendation of better nutritional habits. People with poor diets are more prone to chronic diseases and, in the long term, this can lead to dead. This study proposes a model for the recommendation of nutritional plans using the decision tree technique considering the patient data, in complement with the BMI (Body Mass Index) and BMR (Basal Metabolic Rate) to evaluate and recommend the best nutritional plan for the patient. The algorithm used in the model was trained with a dataset of meal plan data assigned by specialists which were obtained from the Peruvian food composition table, and the data from the diets that were assigned and collected from the nutrition area of the Hospital Marino Molina Sccipa in Lima, Peru. Preliminary results of the experiment with the proposed algorithm show an accuracy of 78.95% allowing to provide accurate recommendations from a considerable amount of historical data. In a matter of seconds, these results were obtained using Scikit learn library. Finally, the accuracy of the algorithm has been proven, generating the necessary knowledge so that it can be used to create appropriate nutritional plans for patients and to improve the process of creating plans for the nutritionist.