Application of latent variable models to autoimmune diseases clinimetry

Abstract: Clinical manifestations of autoimmune diseases often are presented as presence or absence of certain symptoms, which in the clinical context can be associated with disease severity. Classical measures like number of symptoms present have been used to assess the severity of disease, neverth...

Full description

Autores:: Molano González, Nicolás

Tipo de recurso:

Fecha de publicación:: 2015

Institución:: Universidad Nacional de Colombia

Repositorio:: Universidad Nacional de Colombia

Idioma:: spa

Description
Summary:	Abstract: Clinical manifestations of autoimmune diseases often are presented as presence or absence of certain symptoms, which in the clinical context can be associated with disease severity. Classical measures like number of symptoms present have been used to assess the severity of disease, nevertheless this score does not reflect particular aspects of each symptom in the severity manifestation. In this setting latent variable models are a natural choice to address the problem of characterize the disease severity in function of the presence or absence of symptoms in a given patient. On the other hand, we seek associations of this latent trait (i.e severity) with other clinical and demographical information, in order to identify risk factors which can be used in the future for personalized diagnostic and treatment. Having this in mind we investigate extension to the 2pl IRT models, as proposed by Ogasawara (1995), where the individuals latent trait can be modeled by person covariates, and as result of our research, we propose a novel Bayesian algorithm based on working variables for the classical 1pl, 2pl an the extended models involving subject level covariates. This novel algorithm is based on the works of Gamerman (1997), Cepeda-Cuervo (2001), Cepeda-Cuervo and Achcar (2010) and Gutierrez (2015)

Application of latent variable models to autoimmune diseases clinimetry

Publicaciones similares