Catálogo de publicaciones - libros

The QRS complex is the main wave of the ECG. It is widely used for diagnosing many cardiac diseases. Automatic QRS detection is an essential task of cardiac monitoring and many detection algorithms have been proposed in the literature. Although most of the algorithms perform satisfactorily in normal situations, there are contexts, in the presence of noise or a specific pathology, where one algorithm performs better than the others. We propose a combination method that selects, on line, the detector that is the most adapted to the current context. The selection is done by a decision tree that has been learnt from the performance measures of 7 algorithms in various instances of 130 combinations of arrhythmias and noises. The decision tree is compared to expert rules tested in the framework of the cardiac monitoring system .

A public health-care system (HCS) is a complex system that requires permanent monitoring. This paper focuses on the Slovenian national HCS sub-system consisting of a network of health-care professionals at the primary care level. The challenge addressed in this paper is the development and application of intelligent data analysis, decision support and visualization methods aimed to improve the monitoring of human resources of this network. The main outcome is a set of proposed performance indicators and the developed model for monitoring the network of primary health-care professionals of Slovenia. The model enables improved planning and management through data analysis and visualization modules developed for the monitoring of physicians’ qualification, age, workload and dispersion.

This paper describes the application of an information technology infrastructure aimed at supporting translational bioinformatics studies which need the joint management of phenotypic and genotypic data. The system provides an integrated and easy to use software environment, based on data warehouse and data mining tools, to discover the most frequent complex phenotypes and search their penetrance and heritability by mapping them on the population pedigree. We first use a logical formalization to define phenotypes of interest in order to retrieve individuals having that phenotype from the electronic medical record. We then use an open-source Web-based data warehouse application for analyzing phenotypic data and presenting the results in a multidimensional format. Relationships between the selected individuals are automatically visualized by integrating in the system an ad-hoc developed pedigree visualization tool. Finally, the application of the system to support a genetic study of an isolated population, the Val Borbera project, is presented.

The ever-increasing volume of health online information, coupled with the uneven reliability and quality, may have considerable implications for the citizen. In order to address this issue, we propose to use, within a general or specialised search engine, standards for identifying the reliability of online documents. Standards used are those related to the ethics as well as trustworthiness of websites. In this research, they are detected through the URL names of Web pages by applying machine learning algorithms. According to algorithms used and to principles, our straightforward approach shows up to 93% precision and 91% recall. But a few principles remain difficult to recognize.

The cancer disease in Hijuelas-Chile represents the 45% of the population deaths in the last decade. This high mortality rate have concerned the sanitary authority that lacks of information to identify the risk groups and the factors that influence in the disease.

In this work we propose a clustering algorithm for mixed numerical, categorical and multi-valued attributes. We apply our proposed algorithm to identify and to characterize the common patterns in people who died of cancer in the population of Hijuelas between 1994 and 2006. As a consequence of this research, we were able to characterize the people who died of Cancer in Hijuelas-Chile.

A comparative analysis of 14 classic and 23 novel mathematical features for diagnosing lung function in infants is presented. The data set comprises tidal breathing flow volume loops of 195 spontaneously breathing infants aged 3 to 24 months, with 9 known breathing problems (diseases). The data set is sparse. Diagnostic power was evaluated using support vector machines featuring both polynomial and Gaussian kernels in a rigorous experimental setting (100 runs for random splits of data into the training set (90% of data) and test set (10% of data)). Novel features achieve lower error rates than the old ones.

Sarcoidosis is a multisystem disorder that is characterized by the formation of granulomas in certain organs of the body. The exact cause of sarcoidosis is unknown but evidence exists that sarcoidosis results from exposure of genetically susceptible hosts to specific environmental agents. The wide degree of clinical heterogeneity might indicate that sarcoidosis is not a single polymorphic disease but a collection of genetically complex diseases. As a first step to identify the hypothesized subcategories, large amounts of multidimensional data are collected that are divided into distinct levels. We investigated how clustering techniques can be applied to support the interpretation of sarcoidosis and subsequently to reveal categories of sarcoidosis data. An attempt is made to relate multiple clusters between the different data levels based on validation criteria.

Medical reports are predominantly written in natural language; as such they are not computer-accessible. A common way to make medical narrative accessible to automated systems is by assigning ‘computer-understandable’ keywords from a controlled vocabulary. Experts usually perform this task by hand. In this paper, we investigate methods to support or automate this type of medical classification. We report on experiments using the PALGA data set, a collection of 14 million pathological reports, each of which has been classified by a domain expert. We describe methods for automatically categorizing the documents in this data set in an accurate way. In order to evaluate the proposed automatic classification approaches, we compare their output with that of two additional human annotators. While the automatic system performs well in comparison with humans, the inconsistencies within the annotated data constrain the maximum attainable performance.

In this article we present a literature mining method RaJoLink that upgrades Swanson’s ABC model approach to uncovering hidden relations from a set of articles in a given domain. When these relations are interesting from medical point of view and can be verified by medical experts, they represent new pieces of knowledge and can contribute to better understanding of diseases. In our study we analyzed biomedical literature about autism, which is a very complex and not yet sufficiently understood domain. On the basis of word frequency statistics several rare terms were identified with the aim of generating potentially new explanations for the impairments that are observed in the affected population. Calcineurin was discovered as a joint term in the intersection of their corresponding literature. Similarly, NF-kappaB was recognized as a joint term. Pairs of documents that point to potential relations between the identified joint terms and autism were also automatically detected. Expert evaluation confirmed the relevance of these relations.

Intensive care is becoming increasingly complex. If mistakes are to be avoided, there is a need for the large amount of clinical data to be presented effectively to the medical staff. Although the most common approach is to present the data graphically, it has been shown that textual summarisation can lead to improved decision making. As the first step in the BabyTalk project, a prototype is being developed which will generate a textual summary of 45 minutes of continuous physiological signals and discrete events (e.g.: equipment settings and drug administration). Its architecture brings together techniques from the different areas of signal analysis, medical reasoning, and natural language generation. Although the current system is still being improved, it is powerful enough to generate meaningful texts containing the most relevant information. This prototype will be extended to summarize several hours of data and to include clinical interpretation.