Catálogo de publicaciones - libros

The successful completion of the Human Genome Project promised an increase on our knowledge about the way our organism works and therefore would have a major impact in medicine. DNA microarray is one of the techniques that appeared in this “-omic” era and that will certainly change the way diagnosis and disease treatment are made. However, despite the successive scientific breakthroughs the integration of microarrays in clinical practice will face yet the lack of proper information systems and communication standards inside the Health Information Systems (HIS) scenarios. We hereby review current information systems for microarrays’ laboratories and for healthcare institutions and also the latest integration efforts, assessing the shortcomings and structural difficulties derived from integrating two distinct fields. We also present the expected difficulties that may arise from the developments in the genetic diagnosis field and its interactions with other diagnostic areas such as imaging and/or radiology. From this prospective analysis we propose a model where the laboratorial microarray data can be integrated with other diagnostic systems in clinical environments, performing structured diagnostic workflows and integrating information from multiple diagnostic sources onto the HIS.

Bioinformatics analysis of data produced by high-throughput biology, for instance genome projects, is one of the major challenges for the next years. Some of the requirements of this analysis are to access up-to-date databanks (of sequences, patterns, 3D structures, etc.) and relevant algorithms (for sequence similarity, multiple alignment, pattern scanning, etc.). GPS@ is a Web portal devoted to bioinformatics applications on the grid (Grid Protein Sequence Analysis, http://gpsa-pbil.ibcp.fr). GPS@ is the grid release of the NPS@ bioinformatics portal, and is wrapping the mechanisms required for submitting bioinformatics analyses on the grid infrastructure. For example, we have put online two multiple alignment Web Services that are submitting the computing job on a remote grid environment. One is accessible through a classical Web interface by using a simple Web browser; the other one can be used through a SOAP and workflow client such as Taverna or Triana. These Web services can process the submitted alignment on two different computing environments: a local and classical one which is a cluster of 30 CPUs, but we are also providing biologists with a large-scale distributed one: the grid platform of the EU-EGEE project (more than 20,000 CPUs available at the European scale).

The ultimate goal of the biomedical informatics project PrognoChip is the identification of classification and prognosis molecular markers for breast cancer. This requires not only an understanding of the genetic basis of the disease, based on the patient’s tumor gene expression profiles but also the correlation of this data with knowledge normally processed in the clinical setting. In this paper, we present the Mediator component of the PrognoChip Integrated Clinico-Genomics Environment (ICGE), through which the integration of the clinical and genomic information subsystems is achieved. The biomedical investigator can form clinico-genomic queries through the web-based graphical user interface of the Mediator. This is split into several query forms, allowing cancerous sample selection (along with their associated gene expression profiles and patient characteristics), based on criteria of interest. After a query is formed, the Mediator translates it into an equivalent set of local subqueries, which are executed directly against the constituent databases. Then, results are combined for presentation to the user and/or transmission to the Data Mining tools for analysis.

Within the knowledge discovery in databases (KDD) process, previous phases to data mining consume most of the time spent analysing data. Few research efforts have been carried out in theses steps compared to data mining, suggesting that new approaches and tools are needed to support the preparation of data. As regards, we present in this paper a new methodology of ontology-based KDD adopting a federated approach to database integration and retrieval. Within this model, an ontology-based system called OntoDataClean has been developed dealing with instance-level integration and data preprocessing. Within the OntoDataClean development, a preprocessing ontology was built to store the information about the required transformations. Various biomedical experiments were carried out, showing that data have been correctly transformed using the preprocessing ontology. Although OntoDataClean does not cover every possible data transformation, it suggests that ontologies are a suitable mechanism to improve quality in the various steps of KDD processes.

In the paper different methods of construction of language models are discussed in relation to a corpora of medical texts written in an inflective language, namely Polish. The main result is the proposal of a method of language modelling which sequentially combines tri-grams of morphological base forms with tri-grams of words. The introduction of base form tri-grams increased the overall performance of the combined model, measured as the improvement in the accuracy of OCR of handwriting, as well, as the ability to generalisation. The latter was showed by using corpora of two different types as the training one and the test one. The detailed results of tests run on a large corpora of real life medical language are discussed in the paper. An experimental system of OCR of handwritten epicrises utilising the proposed model is presented. The proposed language model decreases the overall error of the system by 64.2% (51% in the case of different types of corpora).

We attempted to find a way to distinguish survivors and non-survivors on the basis of the differences in the dynamics in both patient classes using multivariate autoregressive (MAR) time series analysis techniques. Time series data of 11 physiological variables were used to calculate MAR models. Data were taken from a subset of patients, with an intensive care unit length of stay of at least 20 days, from a database of a previously published randomized controlled trial [1]. The methodology was developed on 20 and validated on 16 patients. Based on the MAR coefficients, impulse response curves were simulated to describe the contributions of a single variable to fluctuations in another. The impulse responses of non-survivors had a tendency to be either more instable or to return to the initial level after a longer time than the responses of survivors did. This allowed us to distinguish survivors from non-survivors in the development cohort with a sensitivity of 0.70 and a selectivity of 1.00. This result was confirmed in the validation set where a sensitivity of 0.63 and a selectivity of 1.00 were reached.

Statistical discrimination of states in the preictal EEG is attempted using a large number of measures from linear and nonlinear time series analysis. The measures are organized in two categories: correlation measures, such as autocorrelation and mutual information at specific lags and new measures derived from oscillations of the EEG time series, such as mean oscillation peak and mean oscillation period. All measures are computed on successive segments of multichannel EEG windows selected from early, intermediate and late preictal states from four epochs. Hypothesis tests applied for each channel and epoch showed good discrimination of the preictal states and allowed for the selection of optimal measures. These optimal measures, together with other standard measures (skewness, kurtosis, largest Lyapunov exponent) formed the feature set for feature-based clustering and the feature-subset selection procedure showed that the best preictal state classification was obtained with the same optimal features.

The analysis of time series databases is very important in the area of medicine. Most of the approaches that address this problem are based on numerical algorithms that calculate distances, clusters, index trees, etc. However, a symbolic rather than numerical analysis is sometimes needed to search for the characteristics of the time series. Symbolic information helps users to efficiently analyse and compare time series in the same or in a similar way as a domain expert would. This paper focuses on the process of transforming numerical time series into a symbolic domain and on the definition of both this domain and a distance for comparing symbolic temporal sequences. The work is applied to the isokinetics domain within an application called I4.

The present work explores the spatiotemporal aspects of the event-related desynchronization (ERD) and synchronization (ERS) during rhythmic finger tapping execution and imagery task. High resolution event related brain potentials were recorded to capture the brain activation underlying the motor execution and motor imagery. ERS and ERD were studied using a complex morlet wavelet decomposition of EEG responses. The results show similar patterns of beta ERD/ERS after the stimulus onset, for both the actual and imagery finger tapping task. This approach and results can be regarded as indicative evidences of a new strategy for recognizing imagined movements in EEG-based brain computer interface research. The long-term objective of this study is to create a multiposition brain controlled switch that is activated by signals that are measured directly from a human’s brain.

Functional Magnetic Resonance Imaging (fMRI) is a non-invasive technique for obtaining a series of images over time under a certain stimulation paradigm. We are interested in identifying regions of brain activity by observing differences in blood magnetism due to haemodynamic response to such stimulus.

Here, we extend Kornak (2000) work by proposing a fully Bayesian two–stage model for detecting brain activity in fMRI. The only assumptions that the model makes about the activated areas is that they emit higher signals in response to an stimulus than non-activated areas do, and that they form connected regions, providing a framework for detecting activity much as a neurologist might.

Due to the model complexity and following the Bayesian paradigm, we use Markov chain Monte Carlo (MCMC) methods to make inference over the parameters. A simulated study is used to check the model applicability and sensitivity.