Evaluación de tecnologías en el cuidado sanitario bajo la perspectiva bayesiana

[ES] La evaluación económica a través de pruebas clínicas forma parte de la práctica clínica de los decisores en salud. Recientemente ha crecido el interés en el uso de la perspectiva bayesiana en el análisis coste-efectividad. En este trabajo repasamos las ventajas de la estadística Bayesiana aplicada a problemas aún en estudio. Así proponemos el uso de covariables en el análisis coste-efectividad para reducir la incertidumbre en la toma de decisiones. Además mostramos el tratamiento de la incertidumbre Bayesiano, así como el uso de más de dos medidas de efectividad para la comparación de tratamientos. En todos los análisis la perspectiva Bayesiana ofrece soluciones innovadoras.

Modelo para la medición con máquinas de visión e iluminación episcópica mediante la aplicación de técnicas de resolución subpíxel y aproximación bayesiana

En esta tesis se desarrolla un modelo físico-matemático, original, que permite simular el comportamiento de las máquinas de visión, en particular las máquinas ópticas digitales, cuando reciben información a través de la luz reflejada por los mensurandos. El modelo desarrollado se lia aplicado para la determinación de los parámetros que intervienen en el proceso de caracterización de formas geométricas básicas, tales como líneas, círculos y elipses. También se analizan las fuentes de error que intervienen a lo largo de la cadena metrológica y se proponen modelos de estimación de las incertidumbres de medida a través un nuevo enfoque basado en estadística bayesiana y resolución subpíxel. La validez del modelo se ha comprobado por comparación de los resultados teóricos, obtenidos a partir de modelos virtuales y simulaciones informáticas, y los reales, obtenidos mediante la realización de medidas de diferentes mensurandos del ámbito electromecánico y de dimensiones submilimétricas. Utilizando el modelo propuesto, es posible caracterizar adecuadamente mensurandos a partir del filtrado, segmentación y tratamiento matemático de las imágenes. El estudio experimental y validación definitiva de los resultados se ha realizado en el Laboratorio de Metrología Dimensional de la Escuela Técnica Superior de Ingeniería y Diseño Industrial de la Universidad Politécnica de Madrid. Los modelos desarrollados se han implementado sobre imágenes obtenidas con la máquina de visión marca TESA, modelo VISIO 300. Abstract In this PhD Thesis an original mathematic-physic model has been developed. It allows simulating the behaviour of the vision measuring machines, in particular the optical digital machines, where they receive information through the light reflected by the measurands. The developed model has been applied to determine the parameters involved in the process of characterization of basic geometrical features such as lines, circles and ellipses. The error sources involved along the metrological chain also are analyzed and new models for estimating measurement uncertainties through a new approach based on Bayesian statistics and subpixel resolution are proposed. The validity of the model has been verified by comparing the theoretical results obtained from virtual models and computer simulations, with actual ones, obtained by measuring of various measurands belonging to the electromechanical field and of submillimeter dimensions. Using the proposed model, it is possible to properly characterize measurands from filtering, segmentation and mathematical processing of images. The experimental study and final validation of the results has been carried out in the "Laboratorio de Metrología Dimensional" (Dimensional Metrology Laboratory) at the Escuela Técnica Superior de Ingeniería y Diseño Industrial (ETSIDI) (School of Engineering and Industrial Design) at Universidad Politécnica de Madrid (UPM). The developed models have been implemented on images obtained with the vision measuring machine of the brand TESA, model VISIO 300.

Generalized factor models: a bayesian approach

There is recent interest in the generalization of classical factor models in which the idiosyncratic factors are assumed to be orthogonal and there are identification restrictions on cross-sectional and time dimensions. In this study, we describe and implement a Bayesian approach to generalized factor models. A flexible framework is developed to determine the variations attributed to common and idiosyncratic factors. We also propose a unique methodology to select the (generalized) factor model that best fits a given set of data. Applying the proposed methodology to the simulated data and the foreign exchange rate data, we provide a comparative analysis between the classical and generalized factor models. We find that when there is a shift from classical to generalized, there are significant changes in the estimates of the structures of the covariance and correlation matrices while there are less dramatic changes in the estimates of the factor loadings and the variation attributed to common factors.

On the density distribution across space: a probabilistic approach

This paper aims at providing a Bayesian parametric framework to tackle the accessibility problem across space in urban theory. Adopting continuous variables in a probabilistic setting we are able to associate with the distribution density to the Kendall's tau index and replicate the general issues related to the role of proximity in a more general context. In addition, by referring to the Beta and Gamma distribution, we are able to introduce a differentiation feature in each spatial unit without incurring in any a-priori definition of territorial units. We are also providing an empirical application of our theoretical setting to study the density distribution of the population across Massachusetts.

An Introduction to parametric and non-parametric models for bivariate positive insurance claim severity distributions

We present a real data set of claims amounts where costs related to damage are recorded separately from those related to medical expenses. Only claims with positive costs are considered here. Two approaches to density estimation are presented: a classical parametric and a semi-parametric method, based on transformation kernel density estimation. We explore the data set with standard univariate methods. We also propose ways to select the bandwidth and transformation parameters in the univariate case based on Bayesian methods. We indicate how to compare the results of alternative methods both looking at the shape of the overall density domain and exploring the density estimates in the right tail.

Modelling dependence in a ratemaking procedure with multivariate Poisson regression models

When actuaries face with the problem of pricing an insurance contract that contains different types of coverage, such as a motor insurance or homeowner's insurance policy, they usually assume that types of claim are independent. However, this assumption may not be realistic: several studies have shown that there is a positive correlation between types of claim. Here we introduce different regression models in order to relax the independence assumption, including zero-inflated models to account for excess of zeros and overdispersion. These models have been largely ignored to multivariate Poisson date, mainly because of their computational di±culties. Bayesian inference based on MCMC helps to solve this problem (and also lets us derive, for several quantities of interest, posterior summaries to account for uncertainty). Finally, these models are applied to an automobile insurance claims database with three different types of claims. We analyse the consequences for pure and loaded premiums when the independence assumption is relaxed by using different multivariate Poisson regression models and their zero-inflated versions.

The Role of model uncertainty and learning in the U.S. postwar policy response to oil prices

This paper studies optimal monetary policy in a framework that explicitly accounts for policymakers' uncertainty about the channels of transmission of oil prices into the economy. More specfically, I examine the robust response to the real price of oil that US monetary authorities would have been recommended to implement in the period 1970 2009; had they used the approach proposed by Cogley and Sargent (2005b) to incorporate model uncertainty and learning into policy decisions. In this context, I investigate the extent to which regulator' changing beliefs over different models of the economy play a role in the policy selection process. The main conclusion of this work is that, in the specific environment under analysis, one of the underlying models dominates the optimal interest rate response to oil prices. This result persists even when alternative assumptions on the model's priors change the pattern of the relative posterior probabilities, and can thus be attributed to the presence of model uncertainty itself.

Policy evaluation and uncertainty about the effects of oil prices on economic activity

This paper addresses the issue of policy evaluation in a context in which policymakers are uncertain about the effects of oil prices on economic performance. I consider models of the economy inspired by Solow (1980), Blanchard and Gali (2007), Kim and Loungani (1992) and Hamilton (1983, 2005), which incorporate different assumptions on the channels through which oil prices have an impact on economic activity. I first study the characteristics of the model space and I analyze the likelihood of the different specifications. I show that the existence of plausible alternative representations of the economy forces the policymaker to face the problem of model uncertainty. Then, I use the Bayesian approach proposed by Brock, Durlauf and West (2003, 2007) and the minimax approach developed by Hansen and Sargent (2008) to integrate this form of uncertainty into policy evaluation. I find that, in the environment under analysis, the standard Taylor rule is outperformed under a number of criteria by alternative simple rules in which policymakers introduce persistence in the policy instrument and respond to changes in the real price of oil.

Industrial Location and Space: New Insights

This paper tries to resolve some of the main shortcomings in the empirical literature of location decisions for new plants, i.e. spatial effects and overdispersion. Spatial effects are omnipresent, being a source of overdispersion in the data as well as a factor shaping the functional relationship between the variables that explain a firm’s location decisions. Using Count Data models, empirical researchers have dealt with overdispersion and excess zeros by developments of the Poisson regression model. This study aims to take this a step further, by adopting Bayesian methods and models in order to tackle the excess of zeros, spatial and non-spatial overdispersion and spatial dependence simultaneously. Data for Catalonia is used and location determinants are analysed to that end. The results show that spatial effects are determinant. Additionally, overdispersion is descomposed into an unstructured iid effect and a spatially structured effect. Keywords: Bayesian Analysis, Spatial Models, Firm Location. JEL Classification: C11, C21, R30.

Compositional ideas in the bayesian analysis of categorical data with application to dose finding clinical trials

Compositional random vectors are fundamental tools in the Bayesian analysis of categorical data.Many of the issues that are discussed with reference to the statistical analysis of compositionaldata have a natural counterpart in the construction of a Bayesian statistical model for categoricaldata.This note builds on the idea of cross-fertilization of the two areas recommended by Aitchison (1986)in his seminal book on compositional data. Particular emphasis is put on the problem of whatparameterization to use

Intrinsic test of independence in contingency tables

A condition needed for testing nested hypotheses from a Bayesianviewpoint is that the prior for the alternative model concentratesmass around the small, or null, model. For testing independencein contingency tables, the intrinsic priors satisfy this requirement.Further, the degree of concentration of the priors is controlled bya discrete parameter m, the training sample size, which plays animportant role in the resulting answer regardless of the samplesize.In this paper we study robustness of the tests of independencein contingency tables with respect to the intrinsic priors withdifferent degree of concentration around the null, and comparewith other “robust” results by Good and Crook. Consistency ofthe intrinsic Bayesian tests is established.We also discuss conditioning issues and sampling schemes,and argue that conditioning should be on either one margin orthe table total, but not on both margins.Examples using real are simulated data are given

Bayesian tools for zero counts in compositional data

The log-ratio methodology makes available powerful tools for analyzing compositionaldata. Nevertheless, the use of this methodology is only possible for those data setswithout null values. Consequently, in those data sets where the zeros are present, aprevious treatment becomes necessary. Last advances in the treatment of compositionalzeros have been centered especially in the zeros of structural nature and in the roundedzeros. These tools do not contemplate the particular case of count compositional datasets with null values. In this work we deal with \count zeros" and we introduce atreatment based on a mixed Bayesian-multiplicative estimation. We use the Dirichletprobability distribution as a prior and we estimate the posterior probabilities. Then weapply a multiplicative modi¯cation for the non-zero values. We present a case studywhere this new methodology is applied.Key words: count data, multiplicative replacement, composition, log-ratio analysis

Econometrics on GPUs

A graphical processing unit (GPU) is a hardware device normally used to manipulate computer memory for the display of images. GPU computing is the practice of using a GPU device for scientific or general purpose computations that are not necessarily related to the display of images. Many problems in econometrics have a structure that allows for successful use of GPU computing. We explore two examples. The first is simple: repeated evaluation of a likelihood function at different parameter values. The second is a more complicated estimator that involves simulation and nonparametric fitting. We find speedups from 1.5 up to 55.4 times, compared to computations done on a single CPU core. These speedups can be obtained with very little expense, energy consumption, and time dedicated to system maintenance, compared to equivalent performance solutions using CPUs. Code for the examples is provided.