Exploration Of Large Data Sets Research Articles

Development and Validation of a Model to Identify Critical Brain Injuries Using Natural Language Processing of Text Computed Tomography Reports

Clinical text reports from head computed tomography (CT) represent rich, incompletely utilized information regarding acute brain injuries and neurologic outcomes. CT reports are unstructured; thus, extracting information at scale requires automated natural language processing (NLP). However, designing new NLP algorithms for each individual injury category is an unwieldy proposition. An NLP tool that summarizes all injuries in head CT reports would facilitate exploration of large data sets for clinical significance of neuroradiological findings. To automatically extract acute brain pathological data and their features from head CT reports. This diagnostic study developed a 2-part named entity recognition (NER) NLP model to extract and summarize data on acute brain injuries from head CT reports. The model, termed BrainNERD, extracts and summarizes detailed brain injury information for research applications. Model development included building and comparing 2 NER models using a custom dictionary of terms, including lesion type, location, size, and age, then designing a rule-based decoder using NER outputs to evaluate for the presence or absence of injury subtypes. BrainNERD was evaluated against independent test data sets of manually classified reports, including 2 external validation sets. The model was trained on head CT reports from 1152 patients generated by neuroradiologists at the Yale Acute Brain Injury Biorepository. External validation was conducted using reports from 2 outside institutions. Analyses were conducted from May 2020 to December 2021. Performance of the BrainNERD model was evaluated using precision, recall, and F1 scores based on manually labeled independent test data sets. A total of 1152 patients (mean [SD] age, 67.6 [16.1] years; 586 [52%] men), were included in the training set. NER training using transformer architecture and bidirectional encoder representations from transformers was significantly faster than spaCy. For all metrics, the 10-fold cross-validation performance was 93% to 99%. The final test performance metrics for the NER test data set were 98.82% (95% CI, 98.37%-98.93%) for precision, 98.81% (95% CI, 98.46%-99.06%) for recall, and 98.81% (95% CI, 98.40%-98.94%) for the F score. The expert review comparison metrics were 99.06% (95% CI, 97.89%-99.13%) for precision, 98.10% (95% CI, 97.93%-98.77%) for recall, and 98.57% (95% CI, 97.78%-99.10%) for the F score. The decoder test set metrics were 96.06% (95% CI, 95.01%-97.16%) for precision, 96.42% (95% CI, 94.50%-97.87%) for recall, and 96.18% (95% CI, 95.151%-97.16%) for the F score. Performance in external institution report validation including 1053 head CR reports was greater than 96%. These findings suggest that the BrainNERD model accurately extracted acute brain injury terms and their properties from head CT text reports. This freely available new tool could advance clinical research by integrating information in easily gathered head CT reports to expand knowledge of acute brain injury radiographic phenotypes.

Demonstrating the voice-based exploration of large data sets with CiceroDB-zero

This demonstration enables participants to explore large data sets via voice interfaces. The focus of the demonstration is on methods generating concise speech descriptions of query results, specified by users via voice input. The technical novelty of the demonstrated system lies in the fact that processing overheads are mostly moved into a pre-processing phase, generating speeches for batches of queries defined via templates. Visitors can access the demo via smart speakers on-site or via their own smart phones. They will be able to customize the generated voice descriptions and to tune voice output methods.

On Evaluating Runtime Performance of Interactive Visualizations.

As our field matures, evaluation of visualization techniques has extended from reporting runtime performance to studying user behavior. Consequently, many methodologies and best practices for user studies have evolved. While maintaining interactivity continues to be crucial for the exploration of large data sets, no similar methodological foundation for evaluating runtime performance has been developed. Our analysis of 50 recent visualization papers on new or improved techniques for rendering volumes or particles indicates that only a very limited set of parameters like different data sets, camera paths, viewport sizes, and GPUs are investigated, which make comparison with other techniques or generalization to other parameter ranges at least questionable. To derive a deeper understanding of qualitative runtime behavior and quantitative parameter dependencies, we developed a framework for the most exhaustive performance evaluation of volume and particle visualization techniques that we are aware of, including millions of measurements on ten different GPUs. This paper reports on our insights from statistical analysis of this data, discussing independent and linear parameter behavior and non-obvious effects. We give recommendations for best practices when evaluating runtime performance of scientific visualization applications, which can serve as a starting point for more elaborate models of performance quantification.

Classifying galaxy spectra at 0.5 < z < 1 with self-organizing maps

The spectrum of a galaxy contains information about its physical properties. Classifying spectra using templates helps to elucidate the nature of a galaxy’s energy sources. In this paper, we investigate the use of self-organizing maps in classifying galaxy spectra against templates. We trained semi-supervised self-organizing map networks using a set of templates covering the wavelength range from far ultraviolet to near-infrared. The trained networks were used to classify the spectra of a sample of 142 galaxies with 0.5 < |$z$| < 1 and the results compared to classifications performed using K-means clustering, a supervised neural network, and chi-squared minimization. Spectra corresponding to quiescent galaxies were more likely to be classified similarly by all methods while starburst spectra showed more variability. Compared to classification using chi-squared minimization or the supervised neural network, the galaxies classed together by the self-organizing map had more similar spectra. The class ordering provided by the 1D self-organizing maps corresponds to an ordering in physical properties, a potentially important feature for the exploration of large data sets.

A Survey on Data Visualization Applications Challenges Techniques and Technologies

Trillions of dollars are spent each year on health care.The Department of Health Research (DHR) keeps track of avariety of health care indicators across the country, resulting in a large geospatially multivariate data set. This paper presents the various techniques, tools, technology and algorithms for the representation of large scale data which aims to provide good overviews of complete structures and the content of the data in one display space. The ability to visualize multiple variables on the map and compare them using a table and charts at the same time can provide valuable insights which might not be possible to obtain from current tools. . There are large numbers of data visualization techniques which have been developed over the last decade to support the exploration of large data sets. The techniques and tools discussed in this paper are based on the data type to be visualized, the visualization technique and the interaction and distortion technique.

DAMEWARE: A Web Cyberinfrastructure for Astrophysical Data Mining

Astronomy is undergoing a methodological revolution triggered by an unprecedented wealth of complex and accurate data. The new panchromatic, synoptic sky surveys require advanced tools for discovering patterns and trends hidden behind data which are both complex and of high dimensionality. We present DAMEWARE (DAta Mining & Exploration Web Application REsource): a general purpose, web-based, distributed data mining environment developed for the exploration of large data sets, and finely tuned for astronomical applications. By means of graphical user interfaces, it allows the user to perform classification, regression, or clustering tasks with machine learning methods. Salient features of DAMEWARE include its ability to work on large datasets with minimal human intervention, and to deal with a wide variety of real problems such as the classification of globular clusters in the galaxy NGC1399; the evaluation of photometric redshifts; and, finally, the identification of candidate Active Galactic Nuclei in multiband photometric surveys. In all these applications, DAMEWARE allowed us to achieve better results than those attained with more traditional methods. With the aim of providing potential users with all needed information, in this paper we briefly describe the technological background of DAMEWARE, give a short introduction to some relevant aspects of data mining, followed by a summary of some science cases and, finally, provide a detailed description of a template use case.

BioNetLink - an architecture for working with network data.

The visualization of biological data gained increasing importance in the last years. There is a large number of methods and software tools available that visualize biological data including the combination of measured experimental data and biological networks. With growing size of networks their handling and exploration becomes a challenging task for the user. In addition, scientists also have an interest in not just investigating a single kind of network, but on the combination of different types of networks, such as metabolic, gene regulatory and protein interaction networks. Therefore, fast access, abstract and dynamic views, and intuitive exploratory methods should be provided to search and extract information from the networks. This paper will introduce a conceptual framework for handling and combining multiple network sources that enables abstract viewing and exploration of large data sets including additional experimental data. It will introduce a three-tier structure that links network data to multiple network views, discuss a proof of concept implementation, and shows a specific visualization method for combining metabolic and gene regulatory networks in an example.

BioNetLink - An Architecture for Working with Network Data

Summary The visualization of biological data gained increasing importance in the last years. There is a large number of methods and software tools available that visualize biological data including the combination of measured experimental data and biological networks. With growing size of networks their handling and exploration becomes a challenging task for the user. In addition, scientists also have an interest in not just investigating a single kind of network, but on the combination of different types of networks, such as metabolic, gene regulatory and protein interaction networks. Therefore, fast access, abstract and dynamic views, and intuitive exploratory methods should be provided to search and extract information from the networks. This paper will introduce a conceptual framework for handling and combining multiple network sources that enables abstract viewing and exploration of large data sets including additional experimental data. It will introduce a three-tier structure that links network data to multiple network views, discuss a proof of concept implementation, and shows a specific visualization method for combining metabolic and gene regulatory networks in an example.

A novel “Integrated Biomarker Response” calculation based on reference deviation concept

Multi-biomarker approaches are used to assess ecosystem health and identify impacts of environmental stress on organisms. However, exploration of large datasets by environmental managers represents a major challenge for regulatory application of this tool. Several integrative tools were developed to summarize biomarker responses. The aim of the present paper is to update calculation of the "Integrated Biological Response" (IBR) described by Beliaeff and Burgeot (Environ Toxicol Chem 21:1316-1322, 2002) to avoid weaknesses of this integrative tool. In the present paper, a novel index named "Integrated Biological Responses version 2" based on the reference deviation concept is presented. It allows a clear discrimination of sampling sites as for the IBR, but several differences are observed for contaminated sites according to up- and downregulation of biomarker responses. This novel tool could be used to integrate multi-biomarker responses not only in large-scale monitoring but also in upstream/downstream investigations.

Pedimap: Software for the Visualization of Genetic and Phenotypic Data in Pedigrees

Pedimap is a user-friendly software tool for visualizing phenotypic and genotypic data for related individuals linked in pedigrees. Genetic data can include marker scores, Identity-by-Descent probabilities, and marker linkage map positions, allowing the visualization of haplotypes through lineages. The pedigrees can accommodate all types of inheritance, including selfing, cloning, and repeated backcrossing, and all ploidy levels are supported. Visual association of the genetic data with phenotypic data simplifies the exploration of large data sets, thereby improving breeding decision making. Data are imported from text files; in addition data exchange with other software packages (FlexQTLTM and GenomeStudioTM) is possible. Instructions for use and an executable version compatible with the Windows platform are available for free from http://www.plantbreeding.wur.nl/UK/software_pedimap.html.

BioMercator V3: an upgrade of genetic map compilation and quantitative trait loci meta-analysis algorithms

Summary: Compilation of genetic maps combined to quantitative trait loci (QTL) meta-analysis has proven to be a powerful approach contributing to the identification of candidate genes underlying quantitative traits. BioMercator was the first software offering a complete set of algorithms and visualization tool covering all steps required to perform QTL meta-analysis. Despite several limitations, the software is still widely used. We developed a new version proposing additional up to date methods and improving graphical representation and exploration of large datasets.Availability and implementation: BioMercator V3 is implemented in JAVA and freely available (http://moulon.inra.fr/biomercator)Contact: joets@moulon.inra.fr

Recursive partitioning

AbstractRecursive partitioning (RP) is a predictive approach with minimal statistical or model assumptions, which models the relationship in terms of trees or dendrograms. It is particularly appropriate for initial exploration of large data sets, especially messy ones, and may either validate other approaches with stronger model assumptions or lead to a final analysis in its own right. It may be used for either a nominal scale (categorical) or an interval‐scale (numeric) dependent variable. A major issue is the size of tree to be fitted; different approaches for this have been proposed. Like many other feature selection methods, RP is unstable, the model being potentially sensitive to minor perturbations in the data. Fitting multiple trees helps explore alternative models, and also provides better predictions than those giving by a single tree. RP is well supported in both commercial and public domain software. Copyright © 2009 John Wiley &amp; Sons, Inc.This article is categorized under: Statistical and Graphical Methods of Data Analysis &gt; Data Reduction, Smoothing, and Filtering

Value and Relation Display: Interactive Visual Exploration of Large Data Sets with Hundreds of Dimensions

Few existing visualization systems can handle large data sets with hundreds of dimensions, since high-dimensional data sets cause clutter on the display and large response time in interactive exploration. In this paper, we present a significantly improved multidimensional visualization approach named Value and Relation (VaR) display that allows users to effectively and efficiently explore large data sets with several hundred dimensions. In the VaR display, data values and dimension relationships are explicitly visualized in the same display by using dimension glyphs to explicitly represent values in dimensions and glyph layout to explicitly convey dimension relationships. In particular, pixel-oriented techniques and density-based scatterplots are used to create dimension glyphs to convey values. Multidimensional scaling, Jigsaw map hierarchy visualization techniques, and an animation metaphor named Rainfall are used to convey relationships among dimensions. A rich set of interaction tools has been provided to allow users to interactively detect patterns of interest in the VaR display. A prototype of the VaR display has been fully implemented. The case studies presented in this paper show how the prototype supports interactive exploration of data sets of several hundred dimensions. A user study evaluating the prototype is also reported in this paper.

An Automated Approach for the Optimization of Pixel-Based Visualizations

During the last two decades, a wide variety of advanced methods for the visual exploration of large data sets have been proposed. For most of these techniques user interaction has become a crucial element, since there are many situations in which users or analysts have to select the right parameter settings from among many in order to construct insightful visualizations. The right choice of input parameters is essential, since suboptimal parameter settings or the investigation of irrelevant data dimensions make the exploration process more time consuming and may result in wrong conclusions. But finding the right parameters is often a tedious process and it becomes almost impossible for an analyst to find an optimal parameter setting manually because of the volume and complexity of today's data sets. Therefore, we propose a novel approach for automatically determining meaningful parameter- and attribute settings based on the combined analysis of the data space and the resulting visualizations with respect to a given task. Our technique automatically analyzes pixel images resulting from visualizations created from diverse parameter mappings and ranks them according to the potential value for the user. This allows a more effective and more efficient visual data analysis process, since the attribute/parameter space is reduced to meaningful selections and thus the analyst obtains faster insight into the data. Real-world applications are provided to show the benefit of the proposed approach.

The future of pediatric nephrology

The delineation of renal disease in children dates to the 1880s with descriptions of Henoch's purpura, bladder exstrophy, renal rickets and nephritis. The discipline of pediatric nephrology mainly emerged during the 20th century in response to problems in fluid and electrolyte balance, characterization of the nephrotic syndrome, use of renal biopsy, antibiotic therapy of urinary tract infections, dialysis and transplantation in children, growth problems associated with chronic renal failure, detection and therapy of hypertension, and the creation of both national and international pediatric nephrology societies and a journal now in its 18th year. The development of molecular and cell biology, genetic and genomic techniques and bioinformatics methods underlie many future directions. We should anticipate further elucidation of single gene disorders, of complex trait analysis of disorders, such as diabetic nephropathy and hypertension, the interplay of developmental genes and gene products and interactions within the podocyte. Specific therapies directed against inflammation, vascular damage, cyst development, the ravages of proteinuria and graft rejection (or induction leading to tolerance) will be developed. Stem cell therapies may replace lost renal mass, even of specific nephron sites. Novel therapies will also modulate the cell cycle, tyrosine kinase signaling and apoptosis. In addition, drugs will be specifically tested in children for many renal conditions. Larger and more specialized registries will be developed; epidemiologic studies and exploration of large data sets will lead to clinical guidelines that are evidenced-based. There is a need for more careful measurement of glomerular filtration rate (GFR), proteinuria and cytokines, and a fuller appreciation of the nutritional and hormonal role of the kidney. Finally, the antecedents of adult renal disease and the need to intervene in a proactive fashion will be realized. Despite these impressive advances in care, the greatest challenges will be in providing children with renal disease access to well-trained pediatric nephrologists, especially in Asia (1 billion children), Africa, Central and South America, and in immigrant and refugee populations. Included in this challenge is the capacity to have affordable access to use of contemporary techniques, and effective medications and prevention strategies. The International Pediatric Nephrology Association (IPNA), its journal, and pediatric advocates will need to use their energies to take on these challenges.

Scalable Visual Data Exploration of Large Data Sets via MultiResolution

During the last decade Visual Exploration and Visual Data Mining tech- niques have proven to be of high value in exploratory data analysis since they combine human visual perception and recognition capabilities with the enormous storage ca- pacity and the computational power of today's computer systems in order to detect patterns and trends in the data. But the ever increasing mass of information leads to new challenges on visualization techniques and concepts. Due to technological progress in computer power and storage capacity today's scientiflc and commercial applications are capable of generating, storing and processing massive amounts of data. Most exist- ing visualization metaphors and concepts do not scale well on such large data sets as interaction capabilities and visual representations sufier from the massive number of data points. To bridge this gap, Visual Analytics aim to incorporate more intelligent means than to just retrieve and display the data items to fllter the relevant from the non-relevant data. In this context the paper introduces a new approach based on a Multiresolution paradigm to increase the scalability of existing Visual data exploration techniques. The basic idea is to provide relevance driven compact representations of the underlying data set that present the data at difierent granularities. In the visualization step the available display space is then distributed according to the data granularity, to emphasize relevant information. The paper aims at introducing a technical base of Multiresolution visualization and provides an application example that shows the usefulness of the proposed approach.

Pixel based visual data mining of geo-spatial data

In many application domains, data is collected and referenced by its geo-spatial location. Spatial data mining, or the discovery of interesting patterns in such databases, is an important capability in the development of database systems. A noteworthy trend is the increasing size of data sets in common use, such as records of business transactions, environmental data and census demographics. These data sets often contain millions of records, or even far more. This situation creates new challenges in coping with scale. For data mining of large data sets to be effective, it is also important to include humans in the data exploration process and combine their flexibility, creativity, and general knowledge with the enormous storage capacity and computational power of today's computers. Visual data mining applies human visual perception to the exploration of large data sets. Presenting data in an interactive, graphical form often fosters new insights, encouraging the formation and validation of new hypotheses to the end of better problem-solving and gaining deeper domain knowledge. In this paper we give a short overview of visual data mining techniques, especially for analyzing geo-spatial data. We provide examples for effective visualizations of geo-spatial data in important application areas such as consumer analysis and census demographics.

On Improved Projection Techniques to Support Visual Exploration of Multi-Dimensional Data Sets

Projection (or dimensionality reduction) techniques have been used as a means to handling the growing dimensionality of data sets as well as providing a way to visualize information coded into point relationships. Their role is essential in data interpretation and simultaneous use of different projections and their visualizations improve data understanding and increase the level of confidence in the result. For that purpose, projections should be fast to allow multiple views of the same data set. In this work we present a novel fast technique for projecting multi-dimensional data sets into bidimensional (2D) spaces that preserves neighborhood relationships. Additionally, a new technique for improving 2D projections from multi-dimensional data is presented, that helps reduce the inherent loss of information yielded by dimensionality reduction. The results are stimulating and are presented in the form of comparative visualizations against known and new 2D projection techniques. Based on the projection improvement approach presented here, a new metric for quality of projection is also given, that matches well the visual perception of quality. We discuss the implication of using improved projections in visual exploration of large data sets and the role of interaction in visualization of projected subspaces.

Local Exact Particle Tracing on Unstructured Grids

Abstract For analyzing and interpreting results of flow simulations, particle tracing is a well established visualization method. In addition, it is a preliminary step for more advanced techniques such as line integral convolution. For interactive exploration of large data sets, a very efficient and reliable particle tracing method is needed. For wind channel experiments or flight simulations, large unstructured computational grids have become common practice. Traditional approachs, based on numerical integration methods of ordinary differential equations however fail to deliver sufficiently accurate path calculation at the speed required for interactive use. In this paper we extend the local exact approach of Nielson and Jung in such a way that it can be used for interactive particle tracing in large data sets of steady flow simulation experiments. This will be achieved by sophisticated preprocessing using additional memory. For further visual enhancement of the streamline we construct an implicitly defined smooth Bézier curve that is used for ray tracing. This allows us to visualize additional scalar values of the simulation as attributes to the trajectory and enables the display of high‐quality smooth curves without creating any visualization geometry and providing a good impression of the spatial situation at the same time. ACM CSS: I.3.3 Computer Graphics—Line and curve generation; I .3.7 Computer Graphics—Raytracing; G.1.2 Numerical Analysis—Spline and piecewise polynomial approximation

Fish mercury levels in lakes—adjusting for Hg and fish-size covariation

Accurate estimates of lake-specific mercury levels are vital in assessing the environmental impact on the mercury content in fish. The intercepts of lake-specific regressions of Hg concentration in fish vs. fish length provide accurate estimates when there is a prominent Hg and fish-size covariation. Commonly used regression methods, such as analysis of covariance (ANCOVA) and various standardization techniques are less suitable, since they do not completely remove the fish-size covariation when regression slopes are not parallel. Partial least squares (PLS) regression analysis reveals that catchment area and water chemistry have the strongest influence on the Hg level in fish in circumneutral lakes. PLS is a multivariate projection method that allows biased linear regression analysis of multicollinear data. The method is applicable to statistical and visual exploration of large data sets, even if there are more variables than observations. Environmental descriptors have no significant impact on the slopes of linear regressions of the Hg concentration in perch ( Perca fluviatilis L.) vs. fish length, suggesting that the slopes mainly reflect ontogenetic dietary shifts during the perch life span.