Visualizing the large-scale datasets output by HPC resources presents a difficult challenge, as the memory and compute power required become prohibitively expensive for end user systems. Novel view synthesis techniques can address this by producing a small, interactive model of the data, requiring only a set of training images to learn from. While these models allow accessible visualization of large data and complex scenes, they do not provide the interactions needed for scientific volumes, as they do not support interactive selection of transfer functions and lighting parameters. To address this, we introduce Volume Encoding Gaussians (VEG), a 3D Gaussian-based representation for volume visualization that supports arbitrary color and opacity mappings. Unlike prior 3D Gaussian Splatting (3DGS) methods that store color and opacity for each Gaussian, VEG decouple the visual appearance from the data representation by encoding only scalar values, enabling transfer function-agnostic rendering of 3DGS models. To ensure complete scalar field coverage, we introduce an opacity-guided training strategy, using differentiable rendering with multiple transfer functions to optimize our data representation. This allows VEG to preserve fine features across a dataset’s full scalar range while remaining independent of any specific transfer function. Across a diverse set of volume datasets, we demonstrate that our method outperforms the state-of-the-art on transfer functions unseen during training, while requiring a fraction of the memory and training time. 
Index Terms - Volume Rendering, 3D Gaussian Splatting

Gaussian density comparison of VEG models trained using either a linear opacity map, an inverse linear opacity map, or our multiple opacity map method using three steps on the RBL dataset.

Volume Encoding Gaussians: Transfer Function-Agnostic 3D Gaussians for Volume Rendering

Evolving measuring and computing capabilities, along with increasingly complex problems or models, are resulting in a new type of dataset: second-generation ensemble data. Like first-generation ensemble data, these data consist of large-scale, repeated measurements of the same process or phenomenon, and often have a spatial component. Unlike older datasets, they typically require the extraction and aggregation of novel complex features, which may be generated through direct measurements rather than simulations, and appear in a wider range of application domains, including neuroscience. We describe an interactive visual analysis solution for this type of second-generation ensemble data, related to the study and planning of surgical interventions in epilepsy treatment. As part of this solution, we introduce a dynamic community abstraction in conjunction with analysis algorithms for feature extraction and aggregation, registration techniques to correlate and project sample data, and custom visual encodings to support the analysis of conserved network patterns. A quantitative and qualitative evaluation with domain experts at four sites demonstrates the effectiveness of this solution. We discuss this approach and solution in the context of second-generation ensemble data analysis, along with the challenges of working with this type of data.

Analyzing conserved electrical activity patterns in epilepsy patients based on ensemble spatial measurements.

Neurosurgery Network Pattern Analysis with 2nd Generation Ensembles

Normal tissue complication probability (NTCP) modeling using dose–volume histograms (DVHs) is fundamentally challenged by high dimensionality, severe multicollinearity, and substantial overlap between DVH profiles of patients with differing toxicity outcomes, limiting the effectiveness of classical classification approaches. We introduce SC–NTCP, a supervised contrastive learning framework that transforms DVH data into a compact, separable latent representation optimized for predicting osteoradionecrosis (ORN) in head and neck cancer patients. Rather than relying on raw, high-dimensional DVH features, SC–NTCP explicitly maximizes intra-class similarity and inter-class separability within the embedding space, enabling more accurate downstream classification. Using a cohort of head and neck cancer patients, we benchmarked SC–NTCP against logistic regression, support vector machines, multilayer perceptrons, and convolutional neural networks. SC–NTCP demonstrated superior discrimination (AUC = 0.77), improved calibration, and enhanced interpretability via gradient based feature attribution, while the integration of clinical covariates further augmented predictive performance. By addressing the inherent limitations of DVH data, SC–NTCP offers a principled and interpretable approach for robust radiation toxicity prediction, with the potential to inform personalized treatment planning and improve clinical outcomes. 
CCS Concepts 
Computing methodologies - Representation learning; Supervised learning. 
Keywords 
Contrastive Learning, Normal Tissue Complication, Osteoradionecrosis, Dose-Volume Histogram

Dose–volume curves for individual patients, showing mandible dose (Gy) on the x-axis versus relative volume (%) on the y-axis. Curves are colored by ORN outcomes (Negative and Positive)

Advancing Normal Tissue Complication Probability Modeling with Supervised Contrastive Learning for Predicting Osteoradionecrosis

The SAGE Suite - SAGE1, SAGE2, and SAGE3 - translates advances in visualization, cyberinfrastructure, and human-computer interaction into an open, scalable platform that aligns with embodied cognition to support collaborative, spatial reasoning on large displays and personal devices. Over two decades and hundreds of deployed walls worldwide, SAGE has enabled scientists, educators, and students to juxtapose heterogeneous media, sustain shared context, and accelerate sensemaking across the research lifecycle. This paper contributes: (1) a synthesis of the Suite&rsquo;s translational impact across domains - from biology and atmospheric science to disaster management, health care, public outreach and workforce development; (2) a comparative framing of SAGE3 (the Smart Amplified Group Environment) among Computer Supported Cooperative Work and infinite-canvas tools; (3) the design rationale and user experience foundations of SAGE3&rsquo;s &ldquo;spatial thinking operating system,&rdquo; including boards, rooms, wall viewports, and multi-user attention/flow mechanisms; (4) a modular architecture that delivers low-latency synchronization, extensibility via plugins, and privacy-aware deployment; and (5) a paradigm for human - Artificial Intelligence (AI) collaboration that spatializes notebooks and conversational workflows, enabling multi-user, multi-AI interaction grounded in shared visual context. We also surface systemic challenges in recognizing software-as-instrument within academic incentives and document emergent usage patterns spanning synchronous/asynchronous, co-located/distributed work. SAGE3 demonstrates how open, research-driven cyberinfrastructure can couple spatial cognition with collective intelligence to advance scientific collaboration and decision-making. 
Keywords: Cyberinfrastructure - Collaboration - Computer supported cooperative work - Artificial intelligence - Human computer interaction - Embodied cognition

SAGE2 supporting a NASA ENDURANCE project meeting in the CAVE2 Hybrid Reality Environment (2011)

Building Collaborative Intelligence: The Translational Journey of the Smart Amplified Group Environment Across Research, Education, and Practice

Symptom modeling in head and neck cancer is challenged by the complexity of heterogeneous patient data, leading to an interest in deep learning approaches. Although Long Short-Term Memory Networks (LSTMs) have shown great results in patient risk prediction, their low interpretability requires data modelers to collaborate with clinical experts to validate the results. We present L-VISP, a human–machine solution that uses visual analytics for LSTM modeling in clinical research. L-VISP uses custom visual encodings to make multiple LSTM variants interpretable, supporting a full range of analysis, from understanding model operations and evaluating performance to interpreting results in a clinical context. We evaluate L-VISP with data modelers and a clinical oncologist and present the takeaways from this multidisciplinary collaboration. 
Keywords: medical XAI, LSTM modeling, multidisciplinary visual analytics, temporal visualization 
CCS Concepts: Human-centered computing - Scientific visualization; Computing methodologies - Machine learning; Applied computing - Life and medical sciences

Model performance analysis for a custom cohort (clinician and modeller activity).

L-VISP: LSTM Visualization for Interpretable Symptom Prediction in Patient Cohorts

In situations where it is difficult to perform robotic laboratory tasks with physical hardware - such as handling radioactive materials, or collecting imitation learning demonstrations at scale - it is useful to be able to simulate the robotic laboratory environment.  Mixed Reality interfaces enable operators to interact with these simulations in an intuitive and immersive format. In order to achieve quality, accurate control from the operator, it is important that the simulated hardware, environment, and interaction paradigms are as good as they can be. In this work, we perform a survey of existing techniques for robot simulation, robot control, 3D scene reconstruction, and VR user interaction, and propose methodology for examining the most effective combination of methods for remote teleoperation. Further, we examine how to address user-perceived differences between the simulation interface and the physical control interface.

The goal is to create a VR-enabled User Interface that communicates with a robot controller to provide a common method for interacting with remote and simulated robotic laboratories.

Using Virtual Reality for Remote Dexterous Manipulation: An Exploration of Methods

This multicenter (National Lung Screening Trial [NLST]) and catchment population–based (University of Illinois Health [UIH], urban and suburban Cook County) cross-sectional study used participants at risk of lung cancer with available lung computed tomography (CT) imaging and follow-up between the years 2015 and 2024. In all, 53,452 in NLST and 11,654 in UIH were included on the basis of age and tobacco use–based risk factors for lung cancer. Cohorts were used for training and testing of deep and machine learning models using clinical features alone or combined with CT image features (hybrid computer vision). An optimized seven-feature clinical model achieved receiver operating characteristic (ROC)-AUC values ranging from 0.64 to 0.67 in NLST and 0.60 to 0.65 in UIH cohorts across multiple years. Incorporation of imaging features to form a hybrid computer vision model significantly improved ROC-AUC values to 0.78-0.91 in NLST but deteriorated in UIH with ROC-AUC values of 0.68-0.80, attributable to Black participants where ROC-AUC values ranged from 0.63 to 0.72 across multiple years. Retraining the hybrid computer vision model by incorporating Black and other participants from the UIH cohort improved performance with ROC-AUC values of 0.70-0.87 in a held-out UIH test set.

A hybrid computer vision model to predict lung cancer in diverse populations

The Dragonfly network, with its high-radix and low-diameter structure, is a leading interconnect in high-performance computing. A major challenge is workload interference on shared network links. Parallel discrete event simulation (PDES) is commonly used to analyze workload interference. However, high-fidelity PDES is computationally expensive, making it impractical for large-scale or real-time scenarios. Hybrid simulation that incorporates data-driven surrogate models offers a promising alternative, especially for forecasting application runtime, a task complicated by the dynamic behavior of network traffic. We present \ourmodel, a surrogate model that combines graph neural networks (GNNs) and large language models (LLMs) to capture both spatial and temporal patterns from port level router data. \ourmodel outperforms existing statistical and machine learning baselines, enabling accurate runtime prediction and supporting efficient hybrid simulation of Dragonfly networks.