169,906 research outputs found
Bayesian Mendelian Randomization for incomplete pedigree data, and the characterisation of Multiple Sclerosis proteins
Full text linkBefore the GWAS (genome-wide association study) era, many genetic determinants of disease were found via analysis of multiplex pedigrees, that is, by looking for genetic markers that run in families in a similar way as disease. GWAS advent has robbed pedigree analysis
of its luster. Future scientific methodology seesaw might bring pedigree analysis back into the spotlight. After the recent discovery of hundreds of disease-associated variants, interest is focusing on the way these variants affect downstream molecular markers, such as transcripts and protein levels, and on the way the resulting changes in these markers in turn affect disease risk. Statistical methods such as Mendelian Randomization (Katan, 1986), hereafter denoted as MR, represent important tools in this effort. Most MR studies are based on data from unrelated individuals, a notable exception being Brumpton et al. (2019). In the present paper we argue that by enriching these data with data from family-related individuals, a number of difficulties that are encountered in MR can be signicantly attenuated. Motivated by the above considerations, this paper discusses extensions of MR to deal with pedigree data. We adopt the Bayesian MR framework proposed by Berzuini and colleagues (Berzuini et al., 2018), and extend it in various ways to deal with pedigree data
Effectiveness of potent antiretroviral therapy on progression of human immunodeficiency virus: Bayesian modelling and model checking via counterfactual replicates
No full textCausality: Statistical Perspectives and Applications
No full textA state of the art volume on statistical causality Causality: Statistical Perspectives and Applications presents a wide-ranging collection of seminal contributions by renowned experts in the field, providing a thorough treatment of all aspects of statistical causality. It covers the various formalisms in current use, methods for applying them to specific problems, and the special requirements of a range of examples from medicine, biology and economics to political science. This book:Provides a clear account and comparison of formal languages, concepts and models for statistical causality. Add
Bayesian Trio Models for Association in the Presence of Genotyping Errors
No full textErrors in genotyping can greatly affect family-based association studies. If a mendelian inconsistency is detected, the family is usually removed from the analysis. This reduces power, and may introduce bias. In addition, a large proportion of genotyping errors remain undetected, and these also reduce power. We present a Bayesian framework for performing association studies with SNP data on samples of trios consisting of parents with an affected offspring, while allowing for the presence of both detectable and undetectable genotyping errors. This framework also allows for the inclusion of missing genotypes. Associations between the SNP and disease were modelled in terms of the genotypic relative risks. The performances of the analysis methods were investigated under a variety of models for disease association and genotype error, looking at both power to detect association and precision of genotypic relative risk estimates. As expected, power to detect association decreased as genotyping error probability increased. Importantly, however, analyses allowing for genotyping error had similar power to standard analyses when applied to data without genotyping error. Furthermore, allowing for genotyping error yielded relative risk estimates that were approximately unbiased, together with 95% credible intervals giving approximately correct coverage. The methods were also applied to a real dataset: a sample of schizophrenia cases and their parents genotyped at SNPs in the dysbindin gene. The analysis methods presented here require no prior information on the genotyping error probabilities, and may be fitted in WinBUGS. © 2003 Wiley-Liss, Inc
GAMEES: a probabilistic environment for expert systems.
No full textThis paper describes GAMEES (Graphical Modelling Environment for Expert Systems), an interactive graphical environment for building and processing Belief Networks and Influence Diagrams. We review the existing systems designed for analogous purposes, and, after a brief introduction to Belief Networks and Influence Diagrams, we describe the graphical interface, discuss algorithms for probabilistic inference on these networks and illustrate the current implementation of GAMEES. The system has been designed for being integrated within wider expert systems and actually it is part of the Therapy Advisor module within an expert system for the management of anemic patients
Hybrid knowledge-based systems for therapy planning
No full textThe design and development of a knowledge-based system (KBS) for therapy planning may benefit from an epistemological analysis of this generic medical task. We specialized a previously formulated epistemological model of medical reasoning toward therapy planning by defining an appropriate ontology and an inference model. We propose a computational framework for the implementation of such epistemological model. Then, we discuss how to choose the formalisms for knowledge representation. It will become evident that a KBS for therapy planning usually requires more than one formalism. We experimented the combined use of production rules, frames, and probabilistic models, such as influence diagrams. From the computational point of view, the system is based on a blackboard control architecture, where control knowledge and domain knowledge are represented explicitly and separately. The medical domain where the system has been experimented is hematology, more specifically the therapy of anemic patients. Clinical examples from this field provide empirical evidence supporting our claims
Bayesian networks for patient monitoring
No full textWe consider a Bayesian statistical approach to model-based prediction of a future patient's response to a therapy, suitable in a wide range of clinical monitoring applications, especially when the observations made on the pathophysiological process of interest are imprecise and sporadic. Potential areas of application range from the predictive control of drug delivery to the management of chronic diseases. A distinctive characteristic of the proposed method is the capability of learning from a database of past patients, by explicitly modeling inter-subject variability of the unknown model parameters, and at an individual's level, by periodic updating of patient-specific parameter estimates on the basis of the accumulating data. By combining information about the population and information contained in the data of the specific patient we improve patient-specific forecasts. In order to make the proposed methodology operational within knowledge-based systems for patient monitoring, we present a Bayesian network representation of the underlying probabilistic model. Inferences involved in the prediction process can thus be performed via general algorithms for probability propagation on a Bayesian network. As an illustration of the proposed methodology we describe numerical results from an application in the field of cancer therapy
- …
