NCSA Welcomes 2023-24 Fellows October 12, 2023 Announcements Artificial IntelligenceData AnalyticsHealth SciencesIntegrated CyberinfrastructureModeling and SimulationPartnershipsSoftware and ApplicationsVisualization Share this page: Twitter Facebook LinkedIn Email National Center for Supercomputing Applications building. By Andrew Helregel The National Center for Supercomputing Applications has named 12 new fellows for the 2023-24 academic school year. The NCSA Fellows program is a competitive program for faculty and researchers at the University of Illinois Urbana-Champaign, which provides seed funding for new collaborations that include NCSA staff as integral contributors to the project. The new fellows and their projects are: Novel Interface Design with Mixed Reality (MR) to Support Collaborative Visual Diagnosis of Concussion by Clinicians and Artificial Intelligence (AI) Inki Kim portrait Shenlong Wang portrait Adam Cross portrait NCSA Fellows: Inki Kim (Health Care Engineering Systems Center), Shenlong Wang (Computer Science) and Adam Cross (Clinical Informatics) NCSA Collaborators: Jessica Saw (Visual Analytics), Chad Olson (Visual Analytics) and Lijiang Fu (UIX Designer) About the Project: Mild traumatic brain injury, also known as concussion, can cause a wide variety of short and long-term disabilities, known collectively as neurocognitive impairments (NCI). The current assessment tools used for NCI are based on subjective physical exams and symptom reporting, neither very precise nor reliable. This project aims to create the first and objective AI-driven system to screen and diagnose concussions. This system will use a mixed-reality mobile app and cloud-based technology to provide clinicians with a 3D representation of a patient’s neurocognitive abilities/impairment. The envisioned system will transform the diagnostic and management approach for concussions and related psychological/behavioral health conditions. Development and Evaluation of the LittleBeats Visualization Tool to Capture Infant Stress Regulation in the Home Environment Nancy McElwain portrait Mark Hasegawa-Johnson portrait NCSA Fellows: Nancy McElwain (Human Development & Family Studies) and Mark Hasegawa-Johnson (Electrical & Computer Engineering) NCSA Collaborators: Lijang Fu (UIX Designer), Fangyu Zhou (UIX Designer) and Chad Olson (Visual Analytics) About the Project: One in five children in the U.S. experiences a mental, emotional, or behavioral (MEB) disorder. MEBs often emerge early and have tremendous economic and societal costs. To meet the growing demand for home-based interventions, the team aims to develop and test an interactive visualization tool in the form of a consumer-facing app for use by parents and clinicians by leveraging PI’s prior work using the LittleBeats platform to assess infant stress regulation processes. This technology will help scale the LittleBeats platform for wider adoption in early childhood care. Forecasting Metritis Cases in Dairy Cattle Using Deep Learning of Genomic and Management Information Sandra Rodriguez-Zas portrait NCSA Fellow: Sandra Rodriguez-Zas (Animal Sciences) NCSA Collaborators: Volodymyr Kindratenko (CAII), Christina Fliege (Genomics Group) and Weihao Ge (Genomics Group) About the Project: This project proposes to address the limitation of models commonly used to describe the association between dairy traits and hundreds of thousands of single nucleotide polymorphisms or SNPs across multiple management and environmental conditions. The innovation lies in using deep learning methods to uncover genotype-by-environment interactions and non-additive (e.g., epistatic, dominant) modes in addition to additive modes of SNP action on dairy traits and improve the detection of superior livestock. The proposed project will offer insights into the benefits of deep learning approaches in identifying previously uncovered genomic locations influencing traits of economic importance to the U.S. dairy industry, in detecting previously uncharacterized epistatic and genotype-by-environment effects influencing dairy health and production traits, and by providing precise genetic merit estimates for genome-enabled improvement of dairy production. Visualizing Invisible Wireless Signals: Immersive Modeling of Multipath Propagation with Realistic Spatial Characteristics Elahe Soltanaghai portrait NCSA Fellow: Elahe Soltanaghai (Computer Science) NCSA Collaborator: David Bock (Data Analysis and Visualization) About the Project: The backlog of 5G (and beyond) wireless technology development is evidenced by the lack of high-resolution wireless channel models that can scrutinize the complex interaction of wireless signals with the surrounding environment. Most wireless channel simulations and ray-tracing algorithms are still highly simplified and lack the necessary linkage between the physical environment and the simulation data. The team uses immersive visualization technology to make invisible wireless signal propagations visible. The resulting visualization will enable wireless emulators as well as network operators augmented with real-time complex simulation signal behavior with major impacts on network engineering, design, monitoring, research, and education. DeepDISC: Detection, Instance Segmentation, and Classification for Astronomical Surveys with Deep Learning Xin Liu portrait NCSA Fellow: Xin Liu (Astronomy) NCSA Collaborators: Volodymyr Kindratenko (CAII), Shirui Luo (Modeling and Simulation) and Matias Carrasco Kind (Astronomy) About the Project: The next generation of massive astronomical surveys, such as the upcoming Legacy Survey Space and Time, will deliver unprecedented amounts of images through the 2020s and beyond. However, an efficient and robust method is still lacking to detect, deblend, and classify sources for upcoming massive surveys. This project aims to develop a deep learning framework, “DeepDISC,” that will efficiently process images and accurately identify blended galaxies in astronomical surveys, improving science outcomes and aiding various astronomical inquiries, from efficiently detecting transients and solar system objects to the nature of dark matter and dark energy. TheVault: Accelerating Scientific Discovery with Reliable and Reproducible Machine Learning Yongjoo Park portrait NCSA Fellow: Yongjoo Park (Computer Science) NCSA Collaborators: Daniel Lapine (ISL), Matthew Krafczyk (Data Analytics & CAII) and Volodymyr Kindratenko (CAII) About the Project: Machine learning (ML) is driving scientific innovations in every field. Despite its significance, today’s ML systems are fragile due to the inherent, volatile nature of computer memory that may lead to data loss and models. This proposal aims to develop a general check-pointing framework, The Vault, that will eliminate the need for manual recording and can automatically and efficiently save/restore the exact states of an ML system (e.g., training set, models) throughout data preparation, training, and predictions by treating the system as a generic finite state machine. An Automated Virtual Imaging and Analysis Pipeline for Medical Applications Jing Qian portrait NCSA Fellows: Frank Brooks (Bioengineering) and Jing Qian (Mayo Clinic) NCSA Collaborators: Maria Jaromin (HIPO) and Mikolaj Kowalik (Software Applications and Data Laboratory) About the Project: Computed medical imaging is integral to the current standard of care for a multitude of common diseases and conditions. Despite the wealth of image data that results from routine diagnostic imaging, the prognostic potential of these same data is largely unrealized. Due to physical and engineering constraints, real imaging systems do not faithfully represent the true object being imaged. The main objective of this proposal is to employ Clowder to effectively integrate heterogeneous simulation, processing, and analysis packages such that the end user sees only the inputs and outputs of a seamless, automated virtual imaging pipeline. Augmenting the Treatments for Degenerative Eye Diseases using AI Yogatheesan Varatharajah portrait NCSA Fellow: Yogatheesan Varatharajah (Bioengineering) NCSA Collaborators: Volodymyr Kindratenko (CAII), Maria Jaromin (HIPO) and Shirui Luo (Data Analytics & CAII) About the Project: This project aims to develop artificial intelligence-based approaches to enable early diagnostics and prognostics for degenerative eye diseases (DEDs) based on low-cost imaging techniques. Because of the increasing prevalence of DEDs and scarcity of trained ophthalmologists in the U.S., there is a critical need to develop low-cost tools to detect DEDs early and predict future disease progression. The team will focus on developing robust preprocessing techniques to identify good quality images, designing a deep learning approach for the early detection of DEDs onset, and developing a longitudinal generative model to describe retinal disease over time. The results will be validated using data obtained from actual patients with DEDs receiving treatment at the Mayo Clinic.