Understanding how individuals interpret charts is a crucial concern for visual data communication. This imperative has motivated a number of studies, including past work demonstrating that causal priors-a priori belief about causal relationships between concepts-can have significant influences on the perceived strength of variable relationships inferred from visualizations. This paper builds on these previous results, demonstrating that causal priors can also influence the types of patterns that people perceive as the most salient within ambiguous scatterplots that have roughly equal evidence for trend and cluster patterns. Using a mixed-design approach that combines a large-scale online experiment for breadth of findings with an in-person think-aloud study for analytical depth, we investigated how users' interpretations are influenced by the interplay between causal priors and the visualized data patterns. Our analysis suggests two archetypal reasoning behaviors through which people often make their observations: contextualization, in which users accept a visual pattern that aligns with causal priors and use their existing knowledge to enrich interpretation, and rationalization, in which users encounter a pattern that conflicts with causal priors and attempt to explain away the discrepancy by invoking external factors, such as positing confounding variables or data selection bias. These findings provide initial evidence highlighting the critical role of causal priors in shaping high-level visualization comprehension, and introduce a vocabulary for describing how users reason about data that either confirms or challenges prior beliefs of causality.

Providing effective guidance for users has long been an important and challenging task for efficient exploratory visual analytics, especially when selecting variables for visualization in high-dimensional datasets. Correlation is the most widely applied metric for guidance in statistical and analytical tools, however a reliance on correlation may lead users towards false positives when interpreting causal relations in the data. In this work, inspired by prior insights on the benefits of counterfactual visualization in supporting visual causal inference, we propose a novel, simple, and efficient counterfactual guidance method to enhance causal inference performance in guided exploratory analytics based on insights and concerns gathered from expert interviews. Our technique aims to capitalize on the benefits of counterfactual approaches while reducing their complexity for users. We integrated counterfactual guidance into an exploratory visual analytics system, and using a synthetically generated ground-truth causal dataset, conducted a comparative user study and evaluated to what extent counterfactual guidance can help lead users to more precise visual causal inferences. The results suggest that counterfactual guidance improved visual causal inference performance, and also led to different exploratory behaviors compared to correlation-based guidance. Based on these findings, we offer future directions and challenges for incorporating counterfactual guidance to better support exploratory visual analytics.

"Correlation does not imply causation" is a famous mantra in statistical and visual analysis. However, consumers of visualizations often draw causal conclusions when only correlations between variables are shown. In this paper, we investigate factors that contribute to causal relationships users perceive in visualizations. We collected a corpus of concept pairs from variables in widely used datasets and created visualizations that depict varying correlative associations using three typical statistical chart types. We conducted two MTurk studies on (1) preconceived notions on causal relations without charts, and (2) perceived causal relations with charts, for each concept pair. Our results indicate that people make assumptions about causal relationships between pairs of concepts even without seeing any visualized data. Moreover, our results suggest that these assumptions constitute causal priors that, in combination with visualized association, impact how data visualizations are interpreted. The results also suggest that causal priors may lead to over- or under-estimation in perceived causal relations in different circumstances, and that those priors can also impact users' confidence in their causal assessments. In addition, our results align with prior work, indicating that chart type may also affect causal inference. Using data from the studies, we develop a model to capture the interaction between causal priors and visualized associations as they combine to impact a user's perceived causal relations. In addition to reporting the study results and analyses, we provide an open dataset of causal priors for 56 specific concept pairs that can serve as a potential benchmark for future studies. We also suggest remaining challenges and heuristic-based guidelines to help designers improve visualization design choices to better support visual causal inference.

Complex, high-dimensional data is used in a wide range of domains to explore problems and make decisions. Analysis of high-dimensional data, however, is vulnerable to the hidden influence of confounding variables, especially as users apply ad hoc filtering operations to visualize only specific subsets of an entire dataset. Thus, visual data-driven analysis can mislead users and encourage mistaken assumptions about causality or the strength of relationships between features. This work introduces a novel visual approach designed to reveal the presence of confounding variables via counterfactual possibilities during visual data analysis. It is implemented in CoFact, an interactive visualization prototype that determines and visualizes counterfactual subsets to better support user exploration of feature relationships. Using publicly available datasets, we conducted a controlled user study to demonstrate the effectiveness of our approach; the results indicate that users exposed to counterfactual visualizations formed more careful judgments about feature-to-outcome relationships.

Roadway safety, especially in rural areas, is one of the most critical components in transportation planning. In collaboration with North Carolina Department of Transportation (NCDOT), UNC Highway Safety Research Center (HSRC), and DOT Volpe National Transportation Systems Center, UNC Renaissance Computing Institute (RENCI) developed a roadside feature detection solution leveraging multiple convolutional neural networks. The solution used an iterative active learning (AL) computer vision model training pipeline integrated into an AI tool to detect safety features such as guardrails and utility poles in geographically distributed NC rural roads. We utilized transfer learning by adopting the Xception neural network architecture [1] as the feature extraction backbone which was then used in an iterative AL process supported by a web-based annotation tool. The annotation tool not only allowed for the collection of annotations through an iterative AL process for multiple safety features, it also enabled visual analysis and assessment of model prediction performance in the geospatial context. AL techniques were used to direct human annotators to label images that would most effectively improve the model aimed at minimizing the number of required training labels while maximizing the model’s performance. The iterative AL process combined with a common feature extraction backbone allowed fast model inference on millions of images in the AL sampling space. This enabled a rapid transition between AL rounds while also reducing the computing requirements for each round. Model feature extraction weights were then fine-tuned in the last round of AL to obtain the best accuracy. Since only about 2.7% of 2.6 million unlabeled images in the AL sampling space contain guardrails, there is a significant class imbalance problem that must be addressed in our AL sampling strategies for the guardrail classification model. In this paper, we present our AI tool processing pipeline and methodology and discuss our AL results and future work. Our AI tool can be used to detect roadside safety features and be extended to also locate them for assessing roadside hazards.

We developed a haptically-enhanced physics simulation to investigate the effects of haptics on the understanding of conceptual concepts related to forces—specifically those related to buoyancy. We evaluated the effects of haptic force feedback, as well as traditional visual representations of forces, on learning via a between-participant user study. Participants completed a buoyancy assessment before and after interacting with the simulation. Haptics enhanced performance regardless of prior knowledge. However, the combined effect of haptics with visual cues differed based on participant prior knowledge. Participants with high prior knowledge significantly improved performance when given both abstract visual cues and haptic feedback combined. Participants with low prior knowledge significantly improved when given haptic feedback alone, and the combination of haptics with visual cues did not improve performance. Our results suggest that the prior knowledge of users and the visual cues used impact the effectiveness of haptically-enhanced simulations with respect to learning outcomes.

BACKGROUND: Recent advances in tissue clearing techniques, combined with high-speed image acquisition through light sheet microscopy, enable rapid three-dimensional (3D) imaging of biological specimens, such as whole mouse brains, in a matter of hours. Quantitative analysis of such 3D images can help us understand how changes in brain structure lead to differences in behavior or cognition, but distinguishing densely packed features of interest, such as nuclei, from background can be challenging. Recent deep learning-based nuclear segmentation algorithms show great promise for automated segmentation, but require large numbers of accurate manually labeled nuclei as training data. RESULTS: We present Segmentor, an open-source tool for reliable, efficient, and user-friendly manual annotation and refinement of objects (e.g., nuclei) within 3D light sheet microscopy images. Segmentor employs a hybrid 2D-3D approach for visualizing and segmenting objects and contains features for automatic region splitting, designed specifically for streamlining the process of 3D segmentation of nuclei. We show that editing simultaneously in 2D and 3D using Segmentor significantly decreases time spent on manual annotations without affecting accuracy as compared to editing the same set of images with only 2D capabilities. CONCLUSIONS: Segmentor is a tool for increased efficiency of manual annotation and refinement of 3D objects that can be used to train deep learning segmentation algorithms, and is available at https://www.nucleininja.org/ and https://github.com/RENCI/Segmentor .

The collection and visual analysis of large-scale data from complex systems, such as electronic health records or clickstream data, has become increasingly common across a wide range of industries. This type of retrospective visual analysis, however, is prone to a variety of selection bias effects, especially for high-dimensional data where only a subset of dimensions is visualized at any given time. The risk of selection bias is even higher when analysts dynamically apply filters or perform grouping operations during ad hoc analyses. These bias effects threaten the validity and generalizability of insights discovered during visual analysis as the basis for decision making. Past work has focused on bias transparency, helping users understand when selection bias may have occurred. However, countering the effects of selection bias via bias mitigation is typically left for the user to accomplish as a separate process. Dynamic reweighting (DR) is a novel computational approach to selection bias mitigation that helps users craft bias-corrected visualizations. This paper describes the DR workflow, introduces key DR visualization designs, and presents statistical methods that support the DR process. Use cases from the medical domain, as well as findings from domain expert user interviews, are also reported.

The collection of large, complex datasets has become common across a wide variety of domains. Visual analytics tools increasingly play a key role in exploring and answering complex questions about these large datasets. However, many visualizations are not designed to concurrently visualize the large number of dimensions present in complex datasets (e.g. tens of thousands of distinct codes in an electronic health record system). This fact, combined with the ability of many visual analytics systems to enable rapid, ad-hoc specification of groups, or cohorts, of individuals based on a small subset of visualized dimensions, leads to the possibility of introducing selection bias-when the user creates a cohort based on a specified set of dimensions, differences across many other unseen dimensions may also be introduced. These unintended side effects may result in the cohort no longer being representative of the larger population intended to be studied, which can negatively affect the validity of subsequent analyses. We present techniques for selection bias tracking and visualization that can be incorporated into high-dimensional exploratory visual analytics systems, with a focus on medical data with existing data hierarchies. These techniques include: (1) tree-based cohort provenance and visualization, including a user-specified baseline cohort that all other cohorts are compared against, and visual encoding of cohort "drift", which indicates where selection bias may have occurred, and (2) a set of visualizations, including a novel icicle-plot based visualization, to compare in detail the per-dimension differences between the baseline and a user-specified focus cohort. These techniques are integrated into a medical temporal event sequence visual analytics tool. We present example use cases and report findings from domain expert user interviews.

Temporal event data are collected across a broad range of domains, and a variety of visual analytics techniques have been developed to empower analysts working with this form of data. These techniques generally display aggregate statistics computed over sets of event sequences that share common patterns. Such techniques are often hindered, however, by the high-dimensionality of many real-world event sequence datasets which can prevent effective aggregation. A common coping strategy for this challenge is to group event types together prior to visualization, as a pre-process, so that each group can be represented within an analysis as a single event type. However, computing these event groupings as a pre-process also places significant constraints on the analysis. This paper presents a new visual analytics approach for dynamic hierarchical dimension aggregation. The approach leverages a predefined hierarchy of dimensions to computationally quantify the informativeness, with respect to a measure of interest, of alternative levels of grouping within the hierarchy at runtime. This information is then interactively visualized, enabling users to dynamically explore the hierarchy to select the most appropriate level of grouping to use at any individual step within an analysis. Key contributions include an algorithm for interactively determining the most informative set of event groupings for a specific analysis context, and a scented scatter-plus-focus visualization design with an optimization-based layout algorithm that supports interactive hierarchical exploration of alternative event type groupings. We apply these techniques to high-dimensional event sequence data from the medical domain and report findings from domain expert interviews.

OBJECTIVE: This article reports results from a systematic literature review related to the evaluation of data visualizations and visual analytics technologies within the health informatics domain. The review aims to (1) characterize the variety of evaluation methods used within the health informatics community and (2) identify best practices. METHODS: A systematic literature review was conducted following PRISMA guidelines. PubMed searches were conducted in February 2017 using search terms representing key concepts of interest: health care settings, visualization, and evaluation. References were also screened for eligibility. Data were extracted from included studies and analyzed using a PICOS framework: Participants, Interventions, Comparators, Outcomes, and Study Design. RESULTS: After screening, 76 publications met the review criteria. Publications varied across all PICOS dimensions. The most common audience was healthcare providers (n = 43), and the most common data gathering methods were direct observation (n = 30) and surveys (n = 27). About half of the publications focused on static, concentrated views of data with visuals (n = 36). Evaluations were heterogeneous regarding setting and measurements used. DISCUSSION: When evaluating data visualizations and visual analytics technologies, a variety of approaches have been used. Usability measures were used most often in early (prototype) implementations, whereas clinical outcomes were most common in evaluations of operationally-deployed systems. These findings suggest opportunities for both (1) expanding evaluation practices, and (2) innovation with respect to evaluation methods for data visualizations and visual analytics technologies across health settings. CONCLUSION: Evaluation approaches are varied. New studies should adopt commonly reported metrics, context-appropriate study designs, and phased evaluation strategies.

OBJECTIVE: This study investigates the use of visualization techniques reported between 1996 and 2013 and evaluates innovative approaches to information visualization of electronic health record (EHR) data for knowledge discovery. METHODS: An electronic literature search was conducted May-July 2013 using MEDLINE and Web of Knowledge, supplemented by citation searching, gray literature searching, and reference list reviews. General search terms were used to assure a comprehensive document search. RESULTS: Beginning with 891 articles, the number of articles was reduced by eliminating 191 duplicates. A matrix was developed for categorizing all abstracts and to assist with determining those to be excluded for review. Eighteen articles were included in the final analysis. DISCUSSION: Several visualization techniques have been extensively researched. The most mature system is LifeLines and its applications as LifeLines2, EventFlow, and LifeFlow. Initially, research focused on records from a single patient and visualization of the complex data related to one patient. Since 2010, the techniques under investigation are for use with large numbers of patient records and events. Most are linear and allow interaction through scaling and zooming to resize. Color, density, and filter techniques are commonly used for visualization. CONCLUSIONS: With the burgeoning increase in the amount of electronic healthcare data, the potential for knowledge discovery is significant if data are managed in innovative and effective ways. We identify challenges discovered by previous EHR visualization research, which will help researchers who seek to design and improve visualization techniques.

In virtual embodiment scenarios the participant in an immersive virtual environment is presented with a first-person view of a virtual body, giving them the illusion that the body is, to some extent, their own. This body-ownership illusion can be strengthened by animating the virtual body based on the user's motion. The sometimes poor head-tracking quality of a full-body tracker can induce simulator sickness, especially when wearing a head-mounted display, so a separate higher-quality head-tracking system is used. We discuss the issues present when integrating the data from two such tracking systems, outline principles for generating appropriate firstperson views that maintain the user's body-ownership illusion, and describe two related methods based on these principles.

We present a novel technique for animating self-avatar eye movements in an immersive virtual environment without the use of eye-tracking hardware, and evaluate our technique via a two-alternative, forced-choice-with-confidence experiment that compares this simulated-eye-tracking condition to a no-eye-tracking condition and a real-eye-tracking condition in which the avatar's eyes were rotated with an eye tracker. Viewing the reflection of a tracked self-avatar is often used in virtual-embodiment scenarios to induce in the participant the illusion that the virtual body of the self-avatar belongs to them, however current tracking methods do not account for the movements of the participants eyes, potentially lessening this body-ownership illusion. The results of our experiment indicate that, although blind to the experimental conditions, participants noticed differences between eye behaviors, and found that the real and simulated conditions represented their behavior better than the no-eye-tracking condition. Additionally, no statistical difference was found when choosing between the real and simulated conditions. These results suggest that adding eye movements to self-avatars produces a subjective increase in self-identification with the avatar due to a more complete representation of the participant's behavior, which may be beneficial for inducing virtual embodiment, and that effective results can be obtained without the need for any specialized eye-tracking hardware.