Please use this identifier to cite or link to this item:
Title: Métodos de aprendizaje automático para la predicción de la distribución anormal de grasa, masa magra y/o masa ósea en individuos infectados por VIH
Author: Ramírez Garrastacho, Manuel
Director: Pérez Álvarez, Nuria
Tutor: Ventura Royo, Carles  
Keywords: machine learning
body composition
Issue Date: 2-Jan-2019
Publisher: Universitat Oberta de Catalunya (UOC)
Abstract: The objective of this work is the development of new prediction models for pathologies related to the distribution of fat mass, lean mass and bone tissue in HIV patients using machine learning methods. Due to the improvement in anti-HIV therapies, patients suffering from this disease have often a life expectancy similar to a healthy person. This fact has led to a growing interest in the treatment of secondary pathologies derived from this syndrome or its treatment. Some of the most frequent secondary pathologies in HIV patients are those related to the distribution of certain tissues, such as lipodystrophy, sarcopenia or osteoporosis and osteopenia. In this project the predictive efficiency of a traditional logistic regression model has been compared with different models created using machine learning. The algorithms used in this work are some of the most used in biomedicine: artificial neural networks, support vector machines and random forest. All of them have been implemented using the statistical language R. The results obtained show that the models generated with these modern algorithms are in most cases as effective as the classical regression model. Sometimes it is even possible to achieve better results using machine learning models. This work shows very promising possibilities in the prediction of the appearance of secondary pathologies in HIV patients using these novel techniques.
Language: Spanish
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
TFM_ManuelRamírez.docx2.58 MBMicrosoft Word XMLView/Open
manurgarrasTFM0119memoria.pdfMemoria del TFM2.43 MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons