FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W20031132> ?p ?o ?g. }

• W20031132 abstract "Diabetes mellitus is a lifelong, incapacitating disease affecting multiple organs. Worldwide prevalence figures estimate that there are 250 million diabetic patients today and that this number will increase by 50% by 2025. The disease is associated with devastating chronic complications including coronary heart disease, stroke and peripheral vascular disease (macrovascular disease) as well as microvascular disorders, leading to damage of kidneys (nephropathy) and eyes (retinopathy). These complications impose an immense burden on the quality of life of the patients and account for more than 10% of health care costs in Europe. Therefore, novel means to prevent the onset and the progression of these devastating diabetic complications are needed.The aim of the work presented in this thesis is to propose novel computational methods to study diabetes complications with a multi-level approach.Diabetes mellitus is a strongly multifactorial disease, and several risks factors (such as genetic, and environmental factors) are combined together in a complex trait, leading to the onset of the disease.Physiological mechanisms that underlie the disease and the onset and progression of the different complications are still mostly unknown.Given the complex nature of diabetes, the study of the complications can be faced with a multi-level modeling approach. In the general scheme for complex disease, such as diabetes, 3 key elements act together to determine the disease status (outcome) of a patient: i) the phenotype, i.e. the set of all metabolic, anthropometric and clinical variables characterizing the patient, ii) the genotype, i.e. the DNA sequence of the patient, iii) the set of interventions on the patient, i.e. therapies and treatments with drugs. All these 3 variables are connected each other through interactions and have a joint effect on the final outcome of the patient.The multi-level approach allows to disjoint the full problem into sub-problems, focusing only on a set of variables and interaction (reflecting a specific level of information) according to available data.In the present work, 3 main levels of study of diabetes complications are considered, and, for each approach, novel methodologies developed during my PhD are proposed.The 3 levels of study considered in the present work are: i) modeling the effect of genotype on the outcome, ii) modeling the effect of phenotype and treatment on the progression of the outcome, iii) modeling the effect of treatment on the phenotype.In the first level of study, diabetes complications are studied from a static point of view, i.e. without considering their progression over time, and the main objective is to identify the genetic biomarkers that allow to predict the disease state of the patients with the final goal to stratify patients according to the risk of developing the disease. Genome Wide Associations Studies (GWAs) are statistical studies aiming at identify those SNPs able to explain the differences observed for a certain outcome (the disease status) between cases (diseased subjects) and controls (healthy subjects) in a study population. Several methods performing univariate and/or multivariate selection have been used in literature for the identification of genetic markers from GWAs data. In this thesis, a novel algorithm for genetic biomarker selection and subjects classification from genome-wide SNP data has been developed. The algorithm is based on the Naive Bayes classification framework, enriched by three main features: i) bootstrap aggregating of an ensemble of Naive Bayes classifiers, ii) a novel strategy for ranking and selecting the attributes used by each classifier in the ensemble, iii) a permutation-based procedure for selecting significant biomarkers, based on their marginal utility in the classification process. The algorithm has been validated on the Wellcome Trust Case-Control Consortium on Type 1 Diabetes and its performance compared with the ones of both a standard Naive Bayes algorithm and HyperLASSO, a penalized logistic regression algorithm from the state-of-the-art in simultaneous genome-wide data analysis.The second level of study is represented by the dynamic analysis of diabetes complications, where the variable “time” plays a major role. In particular, the objective is to model the onset and the progression of diabetes complications over time, using phenotypic and therapeutic information, with the final goal to estimate a probability for the diabetic patient to develop a certain complication, thus optimizing clinical trials and avoiding invasive and expensive tests. So far, several models of diabetes complications are present in literature, but none is able to flexibly integrate accumulating –omics knowledge (i.e. proteomics, metabolomics, genomics) into a clinical macro-level. The most interesting complication models, in fact, are based on Markov Models (also called state transition model) and use phenotypic information to describe the cohort of interest without the possibility to easily integrate additional information. A new in-silico model for simulating the progression of cardiovascular and kidney complications in diabetic patients is presented. The model proposes, as innovative feature, the use of Dynamic Bayesian Networks (DBNs) for modeling the interactions between variables. Compared to Markov Models, which require as many nodes as the number of combinations of variables’ values, DBNs are more advantageous in handling both the structure and possible additional information, since each variable is simply represented by a node in the network. The model was built relying on data from the Diabetes Control and Complications Trial, a multicenter randomized clinical trial designed to compare intensive with conventional therapy with regard to their effects on the development and progression of the early vascular and neurologic. The developed model is able to predict the progression of the main diabetes complications with an accuracy greater than 95% at a population level. The model is suitable to be used as a decision support tool to help clinicians in the therapy design through cost-effectiveness analysis: exploiting the simulations generated through the model, it is possible, for example, to choose the best strategy between two different therapies for treating a specific cohort of patients. To this aim, a user-interface based on the present model is currently under development. The flexible structure of the model will allow to easily add genotypic information in the next feature as a potential mean to improve predictions.The last level of study focuses on the action of a specific drug on a target phenotype, with the final aim to develop rational means to personalize drug therapy and to ensure maximum efficacy with minimal adverse effects. Focusing on cardiovascular diseases as a direct complication of diabetes, aspirin therapy is an important component of cardiovascular prevention for high risk patients. Aspirin performs its preventive action by inhibiting a key enzyme (the prostaglandin-endoperoxide synthase PTGS-1, also known as cyclooxygenase COX-1) in the cascade leading to the production of thromboxane B2 (TxB2), the major factor involved in the platelets aggregation with consequent formation of thrombi. It is known, from literature, that diabetic patients exhibit a different response to aspirin therapy in comparison to healthy subjects, showing a reduced effectiveness of the drug, which is often referred to as ‘aspirin resistance’. Given the lack of a mathematical characterization of these phenomena, the problem was faced using a pharmacodynamics modeling approach, with an explorative intent. Relaying on biological knowledge retrieved from literature, a partially lumped and partially distributed compartmental model was developed, able to describe: i) the kinetics of COX-1 enzyme, from its production within megakaryocytes in bone-marrow to circulating platelets in blood, ii) the pharmacokinetics and pharmacodynamics of aspirin, i.e. its distribution in the body tissues and its interaction with COX-1. The model was tested using data of serum thromboxane TxB2 recovery levels after aspirin withdrawal in healthy subjects. Possible mechanisms to explain the so-called ‘aspirin resistance’ have been finally discussed." @default.
• W20031132 created "2016-06-24" @default.
• W20031132 creator A5001029781 @default.
• W20031132 date "2014-01-30" @default.
• W20031132 modified "2023-09-27" @default.
• W20031132 title "Multi-level modeling and computational approaches to investigate long-term diabetes complications" @default.
• W20031132 hasPublicationYear "2014" @default.
• W20031132 type Work @default.
• W20031132 sameAs 20031132 @default.
• W20031132 citedByCount "0" @default.
• W20031132 crossrefType "journal-article" @default.
• W20031132 hasAuthorship W20031132A5001029781 @default.
• W20031132 hasConcept C121332964 @default.
• W20031132 hasConcept C134018914 @default.
• W20031132 hasConcept C41008148 @default.
• W20031132 hasConcept C555293320 @default.
• W20031132 hasConcept C61797465 @default.
• W20031132 hasConcept C62520636 @default.
• W20031132 hasConcept C71924100 @default.
• W20031132 hasConceptScore W20031132C121332964 @default.
• W20031132 hasConceptScore W20031132C134018914 @default.
• W20031132 hasConceptScore W20031132C41008148 @default.
• W20031132 hasConceptScore W20031132C555293320 @default.
• W20031132 hasConceptScore W20031132C61797465 @default.
• W20031132 hasConceptScore W20031132C62520636 @default.
• W20031132 hasConceptScore W20031132C71924100 @default.
• W20031132 hasLocation W200311321 @default.
• W20031132 hasOpenAccess W20031132 @default.
• W20031132 hasPrimaryLocation W200311321 @default.
• W20031132 hasRelatedWork W1945017255 @default.
• W20031132 hasRelatedWork W2202737263 @default.
• W20031132 hasRelatedWork W2294179723 @default.
• W20031132 hasRelatedWork W2365499955 @default.
• W20031132 hasRelatedWork W2536203301 @default.
• W20031132 hasRelatedWork W2554998089 @default.
• W20031132 hasRelatedWork W2765995311 @default.
• W20031132 hasRelatedWork W2768426515 @default.
• W20031132 hasRelatedWork W2811352279 @default.
• W20031132 hasRelatedWork W2948711235 @default.
• W20031132 hasRelatedWork W3004527153 @default.
• W20031132 hasRelatedWork W3021164475 @default.
• W20031132 hasRelatedWork W3039409118 @default.
• W20031132 hasRelatedWork W3115174901 @default.
• W20031132 hasRelatedWork W3138885962 @default.
• W20031132 hasRelatedWork W3171912624 @default.
• W20031132 hasRelatedWork W3190390315 @default.
• W20031132 hasRelatedWork W3210936026 @default.
• W20031132 hasRelatedWork W7669957 @default.
• W20031132 hasRelatedWork W2788922227 @default.
• W20031132 isParatext "false" @default.
• W20031132 isRetracted "false" @default.
• W20031132 magId "20031132" @default.
• W20031132 workType "article" @default.