Catálogo de publicaciones - libros

We demonstrate the application of a grid infrastructure for conducting text mining over distributed data and computational resources. The approach is based on using LexiQuest Mine, a text mining workbench, in a grid computing environment. We describe our architecture and approach and provide an illustrative example of mining full-text journal articles to create a knowledge base of gene relations. The number of patterns found increased from 0.74 per full-text articles from a corpus of 1000 articles to 0.83 when the corpus contained 5000 articles. However, it was also shown that mining a corpus of 5000 full-text articles took 26 hours on a single computer, whilst the process was completed in less than 2.5 hours on a grid comprising of 20 computers. Thus whilst increasing the size of the corpus improved the efficiency of the text-mining process, a grid infrastructure was required to complete the task in a timely manner.

Human liver cell bioreactors are used in extracorporeal liver support therapy. To optimize bioreactor operation with respect to clinical application an early prediction of the long-term bioreactor culture performance is of interest. Data from 70 liver cell bioreactor runs labeled by low (n=18), medium (n=34) and high (n=18) performance were analyzed by statistical and machine learning methods. 25 variables characterizing donor organ properties, organ preservation, cell isolation and cell inoculation prior to bioreactor operation were analyzed with respect to their importance to bioreactor performance prediction. Results obtained were compared and assessed with respect to their robustness. The inoculated volume of liver cells was found to be the most relevant variable allowing the prediction of low versus medium/high bioreactor performance with an accuracy of 84 %.

The aim of this work is to present a fully “in silico” approach for the identification of genes that might be involved in the susceptibility for non disjunction diseases and their regulation by methylation processes. We have carried out a strategy based on the use of online available bioinformatics databases and programs for the retrieval and identification of interesting genes. As result we have obtained 29 putative susceptibility genes regulated by methylation processes. We were neither on the need of developing new software nor carry out clinical laboratory experiments for the identification of these genes. We consider that this “in silico” methodology is robust enough to provide candidate genes that must be checked “in vivo” due to the clinical relevance of non disjunction diseases with the aim of providing new tools and criteria for their diagnostics.

The increasing availability of automated data collection tools, database technologies and Information and Communication Technologies in biomedicine and health care have led to huge amounts of biomedical and health-care data accumulated in several repositories. Unfortunately, the process of analysis of such data represents a complex task also because data volumes grow exponentially so manual analysis and interpretation become impractical. Fortunately, knowledge discovery in databases (KDD) and data mining (DM) are powerful tools available to medical and research people for help them in explore data and discover useful knowledge. To assess the spread of DM and KDD in biomedicine and health care, we designed and performed a search database of biomedical and health-care scientific literature, for the year interval 1997-2004, and analyzed the obtained results. There has been an increase of application of DM methods in literature of bio-medical informatics research most of which in bioinformatics and genomic area.

Nonnegative Matrix Factorization (NMF) has proven to be a useful tool for the analysis of nonnegative multivariate data. However, it is known not to lead to unique results when applied to nonnegative Blind Source Separation (BSS) problems. In this paper we present first results of an extension to the NMF algorithm which solves the BSS problem when the underlying sources are sufficiently sparse. As the proposed target function has many local minima, we use a genetic algorithm for its minimization.

The concept of the bioprosthesis control system implementation in the dedicated hardware is presented. The complete control algorithm was analysed and the decomposition revealing the parts which could be calculated concurrently was made. Specialized digital circuits providing the wavelet transform and the neural network calculations were designed and successfully verified. The experiment results show that the proposed solution provides the desired dexterity and agility of the artificial hand.

Treatment management in critically ill patients needs to be efficient, as delay in treatment may give rise to deterioration in the patient’s condition. Ventilator-associated pneumonia (VAP) occurs in patients who are mechanically ventilated in intensive care units. As it is quite difficult to diagnose and treat VAP, some form of computer-based decision support might be helpful. As diagnosing and treating disorders in medicine involves reasoning with uncertainty, we have used a Bayesian network as our primary tool for building a decision-support system for the clinical management of VAP. The effects of antibiotics on colonisation with various pathogens and subsequent antibiotic choices in case of VAP were modelled in the Bayesian network using the notion of causal independence. In particular, the conditional probability distribution of the random variable that represents the overall coverage of pathogens by antibiotics was modelled in terms of the conjunctive effect of the seven different pathogens, usually referred to as the . In this paper, we investigate generalisations of the noisy-AND, called . It is shown that they offer a means for further improvement to the performance of the Bayesian network.

The paper presents a modified version of principal component analysis of 3-channel Holter recordings that enables to construct one SVM linear classifier for the selected group of patients with arrhythmias. Our classifier has perfect generalization properties. We studied the discrimination of premature ventricular excitation from normal ones. The high score of correct classification (95%) is due to the orientation of the system of coordinates along the largest eigenvector of the normal heart action of every patient under study.

In a pilot effort to improve health communication we created a method for measuring the familiarity of various medical terms. To obtain term familiarity data, we recruited 21 volunteers who agreed to take medical terminology quizzes containing 68 terms. We then created predictive models for familiarity based on term occurrence in text corpora and reader’s demographics. Although the sample size was small, our preliminary results indicate that predicting the familiarity of medical terms based on an analysis of the frequency in text corpora is feasible. Further, individualized familiarity assessment is feasible when demographic features are included as predictors.

We investigate the dependency between sample size and classification accuracy of three classification techniques: Naïve Bayes, Support Vector Machines and Decision Trees over a set of  8500 text excerpts extracted automatically from narrative reports from the Brigham & Women’s Hospital, Boston, USA. Each excerpt refers to the smoking status of a patient as: current, past, never a smoker or, denies smoking. Our empirical results, consistent with [1], confirm that size of the training set and the classification rate are indeed correlated. Even though these algorithms perform reasonably well with small datasets, as the number of cases increases, both SMV and Decision Trees show a substantial improvement in performance, suggesting a more consistent learning process. Unlike the majority of evaluations, ours were carried out specifically in a medical domain where the limited amount of data is a common occurrence [13][14]. This study is part of the I2B2 project, Core 2.