Clasificación de la densidad mineral ósea utilizando técnicas de aprendizaje automático en niños y adolescentes según edad y sexo

Published in: Industry, Innovation, and Infrastructure for Sustainable Cities and Communities: Proceedings of the 17th LACCEI International Multi-Conference for Engineering, Education and Technology
Date of Conference: July 24-26, 2019
Location of Conference: Montego Bay, Jamaica
Authors: Jose Sulla-Torres (Universidad Católica de Santa María, PE)
Alan Bedoya-Carrillo (Universidad Católica de Santa María, PE)
Rossana Gomez-Campos (Universidad Católica de Santa María, CL)
Marco Cossio-Bolaños (Universidad Católica de Santa María, CL)
(Universidad Católica de Santa María)
Full Paper: #120

Abstract:

Currently, bone health is a field that has become very important as bone-related diseases are becoming more common. Osteoporosis currently causes an estimated 8.9 million fractures annually. Bone mineral density (BMD) and bone mineral content (BMC) is one of the indicators that allows diagnosing the bone health problem. The objective is to classify bone mineral density in children and adolescents using automatic learning techniques. To this end, a descriptive cross-sectional study was developed. We studied school children from 2 educational centers with an age range of 6 to 18 years from the province of Arequipa (Peru). Anthropometric variables were evaluated. Bone mineral density (BMD) and bone mineral content (BMD) were determined. The Body Mass Index (BMI) was calculated, and a comparative study was made of 5 machine learning algorithms related to the subject, which include decision trees, Bayesian networks, decision and regression tables. The result that was obtained determined that the algorithm that best classifies the bone mineral density is that of Random Forest with a percentage of accuracy of 94.87%. This algorithm allowed us to implement software that validated the bone health of schoolchildren between 6 and 18 years, which allows us to conclude that the algorithm obtained can be used in the implementation of prediction software that contributes to the classification and prevention of bone health cases in in children and adolescents.