Search

There is a growing trend of applying machine learning methods to medical datasets in order to predict patients' future status. Although some of these methods achieve high performance, challenges still exist in comparing and evaluating different models through their interpretable information. Such analytics can help clinicians improve evidence-based medical decision making. In this work, we develop a visual analytics system that compares multiple models' prediction criteria and evaluates their consistency. With our system, users can generate knowledge on different models' inner criteria and how confidently we can rely on each model's prediction for a certain patient. Through a case study of a publicly available clinical dataset, we demonstrate the effectiveness of our visual analytics system to assist clinicians and researchers in comparing and quantitatively evaluating different machine learning methods., This is the author's version of the article that has been accepted to PacificVis 2020 Visualization Meets AI Workshop

Machine learning for data-driven diagnosis has been actively studied in medicine to provide better healthcare. Supporting analysis of a patient cohort similar to a patient under treatment is a key task for clinicians to make decisions with high confidence. However, such analysis is not straightforward due to the characteristics of medical records: high dimensionality, irregularity in time, and sparsity. To address this challenge, we introduce a method for similarity calculation of medical records. Our method employs event and sequence embeddings. While we use an autoencoder for the event embedding, we apply its variant with the self-attention mechanism for the sequence embedding. Moreover, in order to better handle the irregularity of data, we enhance the self-attention mechanism with consideration of different time intervals. We have developed a visual analytics system to support comparative studies of patient records. To make a comparison of sequences with different lengths easier, our system incorporates a sequence alignment method. Through its interactive interface, the user can quickly identify patients of interest and conveniently review both the temporal and multivariate aspects of the patient records. We demonstrate the effectiveness of our design and system with case studies using a real-world dataset from the neonatal intensive care unit of UC Davis., Comment: This is the author's version of the article that has been accepted in PacificVis 2020 Visualization Meets AI Workshop

The rapid development of Convolutional Neural Networks (CNNs) in recent years has triggered significant breakthroughs in many machine learning (ML) applications. The ability to understand and compare various CNN models available is thus essential. The conventional approach with visualizing each model's quantitative features, such as classification accuracy and computational complexity, is not sufficient for a deeper understanding and comparison of the behaviors of different models. Moreover, most of the existing tools for assessing CNN behaviors only support comparison between two models and lack the flexibility of customizing the analysis tasks according to user needs. This paper presents a visual analytics system, VAC-CNN (Visual Analytics for Comparing CNNs), that supports the in-depth inspection of a single CNN model as well as comparative studies of two or more models. The ability to compare a larger number of (e.g., tens of) models especially distinguishes our system from previous ones. With a carefully designed model visualization and explaining support, VAC-CNN facilitates a highly interactive workflow that promptly presents both quantitative and qualitative information at each analysis stage. We demonstrate VAC-CNN's effectiveness for assisting novice ML practitioners in evaluating and comparing multiple CNN models through two use cases and one preliminary evaluation study using the image classification tasks on the ImageNet dataset., 12 pages, 6 figures. This manuscript is currently under review

The optimization of water distribution systems (WDSs) is vital to minimize energy costs required for their operations. A principal approach taken by researchers is identifying an optimal scheme for water pump controls through examining computational simulations of WDSs. However, due to a large number of possible control combinations and the complexity of WDS simulations, it remains non-trivial to identify the best pump controls by reviewing the simulation results. To address this problem, we design a visual analytics system that helps understand relationships between simulation inputs and outputs towards better optimization. Our system incorporates interpretable machine learning as well as multiple linked visualizations to capture essential input-output relationships from complex WDS simulations. We demonstrate our system's effectiveness through a practical case study and evaluate its usability through expert reviews. Our results show that our system can lessen the burden of analysis and assist in determining optimal operating schemes., To be included in the IEEE VIS 2021 Short Paper Proceedings

Visual data storytelling is gaining importance as a means of presenting data-driven information or analysis results, especially to the general public. This has resulted in design principles being proposed for data-driven storytelling, and new authoring tools being created to aid such storytelling. However, data analysts typically lack sufficient background in design and storytelling to make effective use of these principles and authoring tools. To assist this process, we present ChartStory for crafting data stories from a collection of user-created charts, using a style akin to comic panels to imply the underlying sequence and logic of data-driven narratives. Our approach is to operationalize established design principles into an advanced pipeline which characterizes charts by their properties and similarity, and recommends ways to partition, layout, and caption story pieces to serve a narrative. ChartStory also augments this pipeline with intuitive user interactions for visual refinement of generated data comics. We extensively and holistically evaluate ChartStory via a trio of studies. We first assess how the tool supports data comic creation in comparison to a manual baseline tool. Data comics from this study are subsequently compared and evaluated to ChartStory's automated recommendations by a team of narrative visualization practitioners. This is followed by a pair of interview studies with data scientists using their own datasets and charts who provide an additional assessment of the system. We find that ChartStory provides cogent recommendations for narrative generation, resulting in data comics that compare favorably to manually-created ones.

Diffusion tensor imaging (DTI) has been used to study the effects of neurodegenerative diseases on neural pathways, which may lead to more reliable and early diagnosis of these diseases as well as a better understanding of how they affect the brain. We introduce a predictive visual analytics system for studying patient groups based on their labeled DTI fiber tract data and corresponding statistics. The systems machine-learning-augmented interface guides the user through an organized and holistic analysis space, including the statistical feature space, the physical space, and the space of patients over different groups. We use a custom machine learning pipeline to help narrow down this large analysis space and then explore it pragmatically through a range of linked visualizations. We conduct several case studies using DTI and T1-weighted images from the research database of Parkinsons Progression Markers Initiative.

A common network analysis task is comparison of two networks to identify unique characteristics in one network with respect to the other. For example, when comparing protein interaction networks derived from normal and cancer tissues, one essential task is to discover protein-protein interactions unique to cancer tissues. However, this task is challenging when the networks contain complex structural (and semantic) relations. To address this problem, we design ContraNA, a visual analytics framework leveraging both the power of machine learning for uncovering unique characteristics in networks and also the effectiveness of visualization for understanding such uniqueness. The basis of ContraNA is cNRL, which integrates two machine learning schemes, network representation learning (NRL) and contrastive learning (CL), to generate a low-dimensional embedding that reveals the uniqueness of one network when compared to another. ContraNA provides an interactive visualization interface to help analyze the uniqueness by relating embedding results and network structures as well as explaining the learned features by cNRL. We demonstrate the usefulness of ContraNA with two case studies using real-world datasets. We also evaluate through a controlled user study with 12 participants on network comparison tasks. The results show that participants were able to both effectively identify unique characteristics from complex networks and interpret the results obtained from cNRL., To appear in IEEE Conference on Visual Analytics Science and Technology (VAST) 2020

International audience; In the context of data visualization and analytics, this report outlines some of the challenges and emerging applications that arise in the Big Data era. In particularly, fourteen distinguished scientists from academia and industry, and diverse related communities, i.e., Information Visualization, Human-Computer Interaction, Machine Learning, Data management & Mining, and Computer Graphics have been invited to express their opinions.

The visualization of a network influences the quality of the mental map that the viewer develops to understand the network. In this study, we investigate the effects of a 3D immersive visualization environment compared to a traditional 2D desktop environment on the comprehension of a network's structure. We compare the two visualization environments using three tasks--interpreting network structure, memorizing a set of nodes, and identifying the structural changes--commonly used for evaluating the quality of a mental map in network visualization. The results show that participants were able to interpret network structure more accurately when viewing the network in an immersive environment, particularly for larger networks. However, we found that 2D visualizations performed better than immersive visualization for tasks that required spatial memory., IEEE Pacific Visualization Symposium 2020

Mind mapping is a popular way to explore a design space in creative thinking exercises, allowing users to form associations between concepts. Yet, most existing digital tools for mind mapping focus on authoring and organization, with little support for addressing the challenges of mind mapping such as stagnation and design fixation. We present Spinneret, a functional approach to aid mind mapping by providing suggestions based on a knowledge graph. Spinneret uses biased random walks to explore the knowledge graph in the neighborhood of an existing concept node in the mind map, and provides "suggestions" for the user to add to the mind map. A comparative study with a baseline mind-mapping tool reveals that participants created more diverse and distinct concepts with Spinneret, and reported that the suggestions inspired them to think of ideas they would otherwise not have explored., ACM CHI 2020

Despite advances in diagnosis and management, cardiovascular diseases (CVDs) remain the leading cause of death in the US. Atherosclerosis is the most common form of CVD and the vulnerability of atherosclerotic plaques to rupture is a primary determinant for risk of catastrophic ischemic events. Current imaging of atherosclerotic disease focuses on assessing plaque size and the degree of luminal stenosis, which are not good predictors of plaque stability. Functional methods to identify biomarkers of inflammation in plaques could facilitate assessment of plaque instability to allow early intervention. In this study, we validate the use of a purpose-built, magnetic resonance imaging (MRI)-compatible positron emission tomography (PET) insert for multimodal, molecular imaging of vulnerable plaques in mice. We illustrate the application of PET to screen for inflamed regions to guide the application of MRI. Molecular MRI visualizes regions of vascular inflammation and is coupled with anatomical MRI to generate detailed maps of the inflammatory marker within the context of an individual vessel. As a testbed for this imaging methodology, we developed a multimodal, iron oxide nanoparticle (NP) targeting vascular cell adhesion molecule-1 (VCAM-1) for simultaneous PET/MRI of vascular inflammation performed on a mouse carotid ligation model. In vitro cell studies confirmed that the NPs are not cytotoxic to liver cells. In vivo simultaneous PET/MRI imaging identified regions of inflammation. Three-dimensional rendering of the MRI data facilitated high-resolution visualization of patterns of inflammation along the injured vessel. Histology validated the co-localization of the NPs with VCAM-1 expression at sites of induced inflammation. The results of this work validate the utility of the simultaneous PET/MR insert as a research tool for small animals and lays groundwork to further advance the potential clinical utility of integrated imaging systems.

This study describes a detailed analysis of museum visitors' decoding process as they used a visualization designed to support exploration of a large, complex dataset. Quantitative and qualitative analyses revealed that it took, on average, 43 seconds for visitors to decode enough of the visualization to see patterns and relationships in the underlying data represented, and 54 seconds to arrive at their first correct data interpretation. Furthermore, visitors decoded throughout and not only upon initial use of the visualization. The study analyzed think-aloud data to identify issues visitors had mapping the visual representations to their intended referents, examine why they occurred, and consider if and how these decoding issues were resolved. The paper also describes how multiple visual encodings both helped and hindered decoding and concludes with implications on the design and adaptation of visualizations for informal science learning venues., IEEE VIS (InfoVis/VAST/SciVis) 2019 ACM 2012 CCS - Human-centered computing, Visualization, Empirical studies in visualization

Dimensionality reduction (DR) is frequently used for analyzing and visualizing high-dimensional data as it provides a good first glance of the data. However, to interpret the DR result for gaining useful insights from the data, it would take additional analysis effort such as identifying clusters and understanding their characteristics. While there are many automatic methods (e.g., density-based clustering methods) to identify clusters, effective methods for understanding a cluster's characteristics are still lacking. A cluster can be mostly characterized by its distribution of feature values. Reviewing the original feature values is not a straightforward task when the number of features is large. To address this challenge, we present a visual analytics method that effectively highlights the essential features of a cluster in a DR result. To extract the essential features, we introduce an enhanced usage of contrastive principal component analysis (cPCA). Our method, called ccPCA (contrasting clusters in PCA), can calculate each feature's relative contribution to the contrast between one cluster and other clusters. With ccPCA, we have created an interactive system including a scalable visualization of clusters' feature contributions. We demonstrate the effectiveness of our method and system with case studies using several publicly available datasets., This is the author's version of the article that has been published in IEEE Transactions on Visualization and Computer Graphics. The final version of this record is available at: 10.1109/TVCG.2019.2934251

Deep Recurrent Neural Network (RNN) has gained popularity in many sequence classification tasks. Beyond predicting a correct class for each data instance, data scientists also want to understand what differentiating factors in the data have contributed to the classification during the learning process. We present a visual analytics approach to facilitate this task by revealing the RNN attention for all data instances, their temporal positions in the sequences, and the attribution of variables at each value level. We demonstrate with real-world datasets that our approach can help data scientists to understand such dynamics in deep RNNs from the training results, hence guiding their modeling process.

Summary: Background: From 1990 to 2016, the mortality of children younger than 5 years decreased by more than half, and there are plentiful data regarding mortality in this age group through which we can track global progress in reducing the under-5 mortality rate. By contrast, little is known on how the mortality risk among older children (5–9 years) and young adolescents (10–14 years) has changed in this time. We aimed to estimate levels and trends in mortality of children aged 5–14 years in 195 countries from 1990 to 2016. Methods: In this analysis of empirical data, we expanded the United Nations Inter-agency Group for Child Mortality Estimation database containing data on children younger than 5 years with 5530 data points regarding children aged 5–14 years. Mortality rates from 1990 to 2016 were obtained from nationally representative birth histories, data on household deaths reported in population censuses, and nationwide systems of civil registration and vital statistics. These data were used in a Bayesian B-spline bias-reduction model to generate smoothed trends with 90% uncertainty intervals, to determine the probability of a child aged 5 years dying before reaching age 15 years. Findings: Globally, the probability of a child dying between the ages 5 years and 15 years was 7·5 deaths (90% uncertainty interval 7·2–8·3) per 1000 children in 2016, which was less than a fifth of the risk of dying between birth and age 5 years, which was 41 deaths (39–44) per 1000 children. The mortality risk in children aged 5–14 years decreased by 51% (46–54) between 1990 and 2016, despite not being specifically targeted by health interventions. The annual number of deaths in this age group decreased from 1·7 million (1·7 million–1·8 million) to 1 million (0·9 million–1·1 million) in 1990–2016. In 1990–2000, mortality rates in children aged 5–14 years decreased faster than among children aged 0–4 years. However, since 2000, mortality rates in children younger than 5 years have decreased faster than mortality rates in children aged 5–14 years. The annual rate of reduction in mortality among children younger than 5 years has been 4·0% (3·6–4·3) since 2000, versus 2·7% (2·3–3·0) in children aged 5–14 years. Older children and young adolescents in sub-Saharan Africa are disproportionately more likely to die than those in other regions; 55% (51–58) of deaths of children of this age occur in sub-Saharan Africa, despite having only 21% of the global population of children aged 5–14 years. In 2016, 98% (98–99) of all deaths of children aged 5–14 years occurred in low-income and middle-income countries, and seven countries alone accounted for more than half of the total number of deaths of these children. Interpretation: Increased efforts are required to accelerate reductions in mortality among older children and to ensure that they benefit from health policies and interventions as much as younger children. Funding: UN Children's Fund, Bill & Melinda Gates Foundation, United States Agency for International Development.

Human’s daily movements exhibit high regularity in a space–time context that typically forms circadian rhythms. Understanding the rhythms for human daily movements is of high interest to a variety of parties from urban planners, transportation analysts, to business strategists. In this paper, we present an interactive visual analytics design for understanding and utilizing data collected from tracking human’s movements. The resulting system identifies and visually presents frequent human movement rhythms to support interactive exploration and analysis of the data over space and time. Case studies using real-world human movement data, including massive urban public transportation data in Singapore and the MIT reality mining dataset, and interviews with transportation researches were conducted to demonstrate the effectiveness and usefulness of our system., Visual Informatics, 1 (2)

The following information is missing from the Funding section: This work is also supported in part by UC Davis RISE program and U.S. National Science Foundation through grants IIS-1528203 and IIS-1320229.

International audience; Node-Link diagrams make it possible to take a quick glance at how nodes (or actors) in a network are connected by edges (or ties). A conventional network diagram of a “contact tree” maps out a root and branches that represent the structure of nodes and edges, often without further specifying leaves or fruits that would have grown from small branches. By furnishing such a network structure with leaves and fruits, we reveal details about “contacts” in our ContactTrees upon which ties and relationships are constructed. Our elegant design employs a bottom-up approach that resembles a recent attempt to understand subjective well-being by means of a series of emotions. Such a bottom-up approach to social-network studies decomposes each tie into a series of interactions or contacts, which can help deepen our understanding of the complexity embedded in a network structure. Unlike previous network visualizations, ContactTrees highlight how relationships form and change based upon interactions among actors, as well as how relationships and networks vary by contact attributes. Based on a botanical tree metaphor, the design is easy to construct and the resulting tree-like visualization can display many properties at both tie and contact levels, thus recapturing a key ingredient missing from conventional techniques of network visualization. We demonstrate ContactTrees using data sets consisting of up to three waves of 3-month contact diaries over the 2004-2012 period, and discuss how this design can be applied to other types of datasets.