University Of California Los Angeles

NSF Awards · FY 2026 · 2026-10

This project develops a rigorous theoretical foundation for multi-operator learning, providing a mathematical framework to understand how neural networks can efficiently learn across collections of complex physical systems. Artificial intelligence (AI) research, and in particular deep learning, has made recent advances in scientific computing, where empirical results outpace our theoretical understanding of why they work and how to design them reliably. This project addresses these questions by establishing theoretical rates and scalings that describe how model size, data, and problem structure govern accuracy and generalization, and by quantifying when a single model can perform effectively across multiple problems. By leveraging structures and mathematical properties, the project aims to identify the mechanisms that enable efficient learning in high-dimensional settings. These advances are important for AI since they provide principles for designing models that are both accurate and resource-efficient, improving predictability, interpretability, and robustness. The project will also contribute to workforce development by training researchers in applied mathematics, applied analysis, and AI techniques. The project establishes a mathematical framework for analyzing approximation and generalization properties of neural networks when applied to collections of nonlinear partial differential equations. The aim is to derive explicit rates for approximation and generalization errors in general prediction settings. The approach will also incorporate structural assumptions to explain gains in sample efficiency and the emergence of algebraic rates in high-dimensional settings. In addition, the project develops theory-driven principles for the design of neural network architectures tailored to multi-operator learning, incorporating data-dependent priors and shared representations. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2026 · 2026-09

The growing reliance on data-driven technologies across industry, science, and society demands a workforce with both strong theoretical foundations and the ability to apply rigorous statistical and mathematical reasoning to real-world problems. This Research Training Group, based in UCLA’s Department of Statistics & Data Science, will train the next generation of researchers in the mathematical and statistical foundations of data science, providing them with the tools to tackle emerging challenges in AI, biotechnology, and applied statistics. The Research Training Group will create an integrated ecosystem of research and training, engaging students at all academic levels, including high school, community college, undergraduate, graduate, and postdoctoral levels. Undergraduate students will participate in course-based research experiences and immersive summer research programs, working alongside graduate students and faculty. Graduate students and postdocs will receive comprehensive mentoring through specialized topic courses, working groups, seminars, workshops, and summer schools. A new seminar series will highlight connections between statistics, data theory, and applications across disciplines. These efforts will help build a mathematically trained workforce ready to engage with the demands of modern data science and its applications. The research project focuses on advancing the mathematical and statistical foundations of modern data science, with an emphasis on theory, methodology, and applications. In deep learning, the project will develop theory for large-scale non-convex optimization, algorithmic regularization, generalization in neural networks, and emerging phenomena such as feature learning. Foundational work on generative AI models will address scaling laws, prompt tuning, and statistical principles for in-context learning and diffusion models. The project also develops rigorous statistical methods to support trustworthy and reliable AI, as well as tools for enhancing the accuracy and interpretability of genomics and single-cell technologies, data integration, and biological network inference. Additional efforts focus on statistical approaches for practical causal inference in observational studies, methodologies for studying hidden populations, epidemic modeling from limited data, and advanced change point detection. Across these areas, the project combines theoretical development with use-inspired applications in AI, biology, medicine, and other sciences. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2026 · 2026-07

When people make decisions under uncertainty or with incomplete information, people may rely on mental “fill‑ins” that are part of everyday intelligence but can sometimes lead to errors, like jumping to the wrong conclusion. This project asks why those “quick but wrong” choices happen and whether they arise from the same basic rules of intelligent behavior in humans and comparative animal models. Examining these processes in species that lack language is important for being able to understand the origins of reasoning and the principles that are independent of language abilities. Investigating how different species fill in missing information has potential for determining core principles of learning and reasoning under conditions of uncertainty. This understanding is important for fundamental brain science and relevant for artificial-intelligence (AI) engineers in building tools that guide smarter, safer decisions. The project also provides hands‑on research and data‑science training for high‑school, undergraduate, and graduate students, helping to prepare the next generation of scientists and AI‑literate STEM professionals. The research focuses on a phenomenon called the positive contingency bias, which is a kind of error made when making probabilistic inferences. The project focuses on a tendency to assume that a hidden part of a familiar scene is still present, even when that guess can be wrong. The team plans to develop matched computer‑based and conditioning tasks for humans and comparative animal models to test how different species infer the state of hidden cues as part of their natural intelligence. Planned experiments examine when the positive contingency bias grows stronger or weaker, whether it can cause learning about cues that are never directly seen (only imagined), and whether separate imagined features can be combined into new mental images that guide decision-making. The data analysis plan involves mixed‑effects and Bayesian approaches to compare learning and decision patterns across species, providing tests of associative‑learning and cognitive theories of reasoning under uncertainty. Finding the same bias in non‑linguistic animals would show that some reasoning “fallacies” and shortcuts are built‑in features of biological intelligence, which is relevant for basic brain science and improving the design of next‑generation AI and augmented‑intelligence systems that must operate safely and adaptively in uncertain real‑world environments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2026 · 2026-07

The Interactive Discovery Laboratory (iDLab) a unified, easy-to-use web-based platform that enables researchers to seamlessly access and use computing and data resources across five NSF-supported high-performance computing (HPC) sites and two public cloud platforms. iDLab provides a consistent catalog of interactive applications, access to local data resources, and a shared data partition accessible across all sites. By mitigating technical barriers, reducing data transfer requirements, and maintaining a consistent application environment, iDLab empowers researchers, broadens participation in advanced computing, and accelerates data-driven and AI-enabled discovery across a wide range of scientific fields, including natural hazards engineering, spatial biology, neuroscience, geospatial analytics, and computational physics. Additionally, the platform enhances STEM education through browser-based instructional environments, supports workforce development via coordinated training and tutorials, and advances open science by promoting research outputs that are findable, accessible, interoperable, and reusable (FAIR), thereby maximizing the impact and reach of federal investments in research infrastructure. Led by the University of California, Los Angeles (UCLA), in collaboration with the National Center for Supercomputing Applications (NCSA), the Pittsburgh Supercomputing Center (PSC), Purdue University, the San Diego Supercomputer Center (SDSC), and the Texas Advanced Computing Center (TACC), iDLab transitions the OneSciencePlace platform from a pilot to a national-scale, data-centric production service spanning NSF-supported HPC resources and two public cloud platforms. iDLab provides a consistent interactive application catalog across all partner sites, dedicated GPU and CPU resources for interactive workloads, and a shared filesystem that enables near-real-time cross-site data access without manual data staging. The full research data lifecycle, from ingestion and interactive analysis through collaborative sharing and FAIR publication is supported by the platform. iDLab complements and extends ACCESS and the National AI Research Resource (NAIRR) by providing the interactive layer for cross-facility data access and rapid scientific exploration. Throughout operations, iDLab continues to evolve through ongoing enhancements to platform, applications, data services, and the user experience, driven by operational experience and community needs. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2026 · 2026-07

Dr. Matthew Nava of the University of California, Los Angeles will develop new ways to transform the basic building blocks of chemicals into valuable products including medicines, agrochemicals, material precursors, and chemical fuels. Currently, many essential chemical reactions require the use of rare and expensive “noble” metals, which can be difficult to extract, and subject to supply chain disruptions when sourced from geopolitically sensitive regions. This project seeks to replace these critical metals with common, inexpensive and earth-abundant elements molecularly engineered to perform at the same high level as their noble metal counterparts. A key enabling feature of this work is the use of supporting ligands to compensate for deficiencies of base elements. By making chemical manufacturing cleaner, more affordable, and hardened against supply chain disruptions, this work would strengthen advanced manufacturing in the U.S., while also increasing our economic competitiveness and security. Beyond the laboratory, the project will provide unique training opportunities for undergraduate students, equipping them with the technical skills needed for the modern workforce and creating publicly available tutorials to make applied chemical synthesis available to everyone. The proposed research will investigate the role of metal-ligand cooperativity in bond activation processes through the development of modular synthetic platforms. Specifically, the project would focus on the synthesis of metal aminoxide and related scaffolds designed as metal-based analogues of organic 1,3-dipoles. These complexes will be used to interrogate the fundamental design parameters required to unlock the activation of strong C–H bonds using abundant main-group elements and first-row transition metals. Through detailed structure-function studies and physical characterization, the research will seek to elucidate the individual electronic contributions of the dipole components, which have been historically difficult to decouple. The project will evaluate the ability of these dipolar systems to enable reversible sp2 and sp3 C–H bond activation processes and subsequent functionalization under mild conditions. Ultimately, these molecular design concepts are expected to be broadly applicable to many elements across the periodic table and provide alternative and orthogonal routes to chemical processes that traditionally require precious metals. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2026 · 2026-07

Human connection is a fundamental need, as essential to health and longevity as diet and exercise, and one of the strongest predictors of economic mobility and well-being. Yet Americans experiencing a profound loneliness epidemic, with wide-ranging consequences for public health, productivity, and social stability. At the very same time, artificial intelligence (AI) is rapidly becoming embedded in everyday American life, with millions of citizens interacting with AI systems for conversation, support, and companionship. This conference brings together leading scholars across psychology, computer science, neuroscience, communication, anthropology, and related fields with applied scientists and figures at the coalface of AI to address urgent issues of broad public importance: how AI is reshaping human connections and human flourishing with translational benefits for individuals and society. Through invited talks, data blitzes, poster sessions, and a facilitated agenda-setting discussion, the conference examines how AI affects the core psychological tasks of friendship. The conference integrates perspectives from basic science, applied research, and technology development to unlock the next generation of human-technology relationship science. In addition to generating new collaborations and training early-career scholars, the conference produces shared research priorities and public-facing products with real-world translational benefits for years to come, positioning behavioral science to play a central role in understanding and shaping America’s health in an AI-driven world. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2026 · 2026-06

Project Summary Chronic wound repair is a critical therapeutic need, particularly for wounds arising from mechanical trauma, burns, surgeries, or chronic conditions such as diabetes. Current wound dressings primarily provide physical protection but often lack the multifunctionality needed to support the body’s natural healing processes, especially in the presence of inflammation or infection. Commercially available bioadhesives face limitations, including multiple synthetic steps, low mechanical strength, inadequate anti-inflammatory properties, safety concerns, and high costs, which hinder their clinical application. To address these limitations, we synthesized UgiGel, an innovative gelatin-based, self-healing, bioadhesive engineered through a one-step multicomponent reaction without the need for a catalyst, light activation, or post-purification. Our design employs a one-pot Ugi four- component reaction (Ugi-4CR), utilizing gelatin (amine source), 4-formylphenylboronic acid (4-FPBA, aldehyde source), gallic acid (GA, carboxylic acid source), and cyclohexyl isocyanide (CyIso) to create a pseudopeptide- functionalized gelatin-based adhesive hydrogel. The unique internal crosslinking mechanism relies on dynamic borate ester bonds between the boronic groups of 4-FPBA and the catechol groups of GA, forming a viscoelastic hydrogel upon slight pH and temperature changes. In our preliminary work, we reported the successful synthesis of multifunctional UgiGel bioadhesive, which can be potentially used as a wound dressing. Although our initial UgiGel formulation demonstrated in vitro biocompatibility and adhesion strength higher than some commercial bioadhesives (Figs 1 and 2 in the RS), it faces several limitations, including insufficient skin-friendly detachability, suboptimal antioxidant activity, and limited ex vivo adhesion to wet tissue. Moreover, the in vivo efficacy of UgiGel as an immunomodulatory wound dressing remains untested. Building on this foundation, in this proposal, we aim to reformulate UgiGel to enhance its adhesion performance compared to leading clinical adhesives, introduce skin-friendly thermoresponsive detachability, and boost its anti-inflammatory and antibacterial properties by tuning the component ratios and crosslinking conditions (Aim 1). Adhesion will be activated upon contact with warm skin, while painless detachment will be achieved via localized cooling at 4°C (ice bag). In vitro bioactivity will be assessed using the scratch assay and a 3D injured skin model, along with antibacterial testing, antioxidant assays, and oxygen diffusion analysis. Finally, in collaboration with Dr. Philip Scumpia (Dermatologist, UCLA), we will evaluate the in vivo wound-sealing and healing potential of optimized UgiGel formulations using two murine models: 1) a linear incisional model to assess wound sealing and tensile strength of wounded tissues vs. unwounded tissues, and 2) a splinted excisional wound model to evaluate regenerative capacity and local immune response (Aim 2). Based on our preliminary data, we believe that successful completion of the proposed study will result in a novel, bioadhesive hydrogel that not only seals wounds effectively but also promotes tissue healing and regeneration, offering a scalable and efficient solution for clinical wound management.

NIH Research Projects · FY 2026 · 2026-06

PROJECT SUMMARY/ABSTRACT Recent updates to US Preventive Services Task Force guidelines now recommend T2D/prediabetes screening for all adults ages 35-70 with overweight/obesity (BMI >25). These guidelines expanded to include younger ages (prior guidelines started at age 40) and specify that individuals who screen positively for prediabetes be referred to evidence-based prevention and wellness programs. Despite the promise offered by updated national guidelines and their potential to improve health outcomes, screening uptake remains stubbornly elusive; furthermore, there is almost no research on how to increase rates of screening for prediabetes and T2D, or rigorous studies testing the effectiveness of messaging strategies to increase T2D/prediabetes screening. In response to this urgent need, we propose the Promoting Uptake of Evidence-Based DiaBetes and Prediabetes Screening to Increase HeaLth Effectiveness Study (PEBBLES) study. PEBBLES will leverage an innovative SMART (Sequential Multiple Assignment Randomization Trial) experiment to test an adaptive and sustainable screening intervention strategy using messages delivered by text, EHR patient portal, or letter. The PEBBLES study has the following specific aims: Specific Aim 1. Conduct a series of focus groups and user testing among diverse populations to develop and co-design outreach messages for various modalities (e.g., text, phone, portal) for T2D/prediabetes screening. Specific Aim 2. Implement a randomized SMART experiment engaging individuals aged 35-70 with overweight or obesity (n=2,250) across two health care delivery settings to test the most effective sequence of message delivery strategies for improving uptake of evidence-based screening for T2D and prediabetes. Specific Aim 3. Guided by the Replicating Effective Practices (REP) and the Consolidated Framework for Implementation Research (CFIR) frameworks, conduct a rigorous implementation evaluation of scalability, sustainability, program cost-effectiveness, and potential unintended consequences (e.g., intervention- generated inequalities for some groups). Our multi-site study team has experience in designing and implementing pragmatic screening and T2D prevention trials in clinical settings. The findings from the PEBBLES study will significantly increase the potential to improve T2D/prediabetes screening rates, prevent T2D, accelerate treatment for newly diagnosed T2D, and promote health equity at the population level. At the conclusion of this project, we expect to have identified actionable evidence to guide approaches for glycemic screening among the millions of US adults with undiagnosed T2D or prediabetes.

NIH Research Projects · FY 2026 · 2026-06

Abstract Clinically non-functioning pituitary tumors (CNFTs) account for one-third of all pituitary adenomas. Unlike functioning lactotroph, somatotroph and corticotroph pituitary tumors, the majority of CNFTs derive from the gonadotroph cell lineage (~70%) and do not typically cause clinical and/or biochemical evidence of tumor- related hormone hypersecretion. Instead, they grow insidiously and present with mass effect symptoms such as headache, visual field deficits and hypopituitarism. Surgical resection is the current standard of care for gonadotroph tumors, but complete resection is infrequently achieved due to their size and invariable invasion of locally adjacent structures. Regrowth of the residual tumor occurs in approximately 50% of patients who then may need repeated surgeries and/or radiotherapy. The latter is quite effective in achieving tumor control but carries a risk of hypopituitarism of ~ 40% at 10 years. There are currently no approved medical treatments for gonadotroph tumors, and there is a clear unmet need for novel safe and effective medical therapies for these comparatively common tumors. In this highly innovative project, we will combine our unique patient- derived 3D gonadotroph tumoroid model with an automated high throughput screen (HTS) to identify novel tumoricidal and growth inhibitory compounds for gonadotroph tumors. Aim 1 of our proposal will quantitate inhibition of cell viability as primary endpoint in our high throughput screen to identify small molecules that specifically and efficiently target gonadotroph tumors. Hits will be defined using robust z-score statistics. Potential hits will then be reassessed by a series of orthogonal assays to corroborate their inhibition of cell viability. Compounds that emerge from these rigorous confirmation evaluations will be considered as primary hits. A second aim will use a cascade of follow-up assays including dose response curves, toxicity profiling, pharmacophore modeling, potency analysis and drug-likeness property evaluation to characterize the primary hit compounds. Thereafter, we will combine in-silicon target prediction with a series of complimentary overlapping experimental approaches to deconvolute the mode of actions of the resultant hits. Transcriptome profiling by RNA-seq will be used to segregate drug target pathways which will then be corroborated by pathway disruption using siRNA, shRNA, or CRISPR sub-libraries in combination with cDNA clones. In summary, we are confident that our rigorous HTS using our unique human gonadotroph tumoroids will enable us to find and characterize potent and efficacious disease-targeted small molecule compounds that can be ready for further study and development.

NIH Research Projects · FY 2026 · 2026-06

ABSTRACT Colorectal cancer, with rectal cancer comprising approximately 1/3 of all colorectal cases, is the third most common cancer diagnosed and the second most common cause of cancer-related death in the United States. In 2023, there was an estimated 46, 050 new cases of rectal cancer, and almost 27,000 deaths, or about 60% of the patients dying from this disease. Additionally, there is an alarming increase in the diagnosis of rectal cancer in younger patients (<50 years old). Until recently, the established standard of care for locally advanced rectal cancer (LARC) involved pre-operative long course chemoradiation (LCRT) followed by total mesorectal excision (TME) and adjuvant chemotherapy. However, not all patients derive equal benefit from this approach, evidenced by a relatively low complete pathologic response (pCR) rate to LCRT alone, ranging from 15-27%. More recent strategies have aimed to reduce recurrence rates by applying total neoadjuvant therapy (TNT) to increase the rates of complete clinical response (cCR), allowing patients to avoid surgery (non-operative management, NOM). To increase the complete response (CR, including cCR and pCR) rate, which correlates with better outcomes, and to explore a more cost-effective approach to NOM, neoadjuvant short-course radiation (SCRT) followed by chemotherapy (TNT) is emerging as a promising strategy, offering greater patient convenience, cost- effectiveness, and efficient resource utilization. Studies have demonstrated that this regimen can achieve a CR rate twice as high as that achieved with LCRT alone. However, SCRT has been associated with a higher failure rate in some instances, supporting the notion that there is no universal solution for all LARC patients. Therefore, there is an urgent need to develop, on a per-individual basis, a reliable method to predict whether LCRT or SCRT will offer the highest likelihood of achieving CR, enabling NOM as well as the highest cure rates, for LARC patients. Our goal is to develop a powerful and clinically ready signature, applying both imaging and a unique class of genetic biomarkers, that will allow physicians and their patients to identify the best personalized treatment approach in LARC, either LCRT or SCRT, as measured by achieving a CR. To achieve this goal, we will apply insights into LARC characterization derived from medical imaging, as well as apply novel patient-specific germline genetic biomarkers. To this end, we will build an interpretable radiomics pipeline, consisting of a CNN feature extractor on multi-modal images, a superior multi-objective feature selection algorithm and a model interpreter. In addition, we will develop predictive genetic signatures of response to LCRT versus SCRT in LARC using a large panel of microRNA-based germline biomarkers we have previously shown predicting radiation response. Finally, we will develop tiered fusion models that combine the image and germline signatures to predict the response likelihood, estimate treatment effects, and investigate individualized treatment rules to suggest the treatment type with the highest response probability, assisting physicians and patients in treatment selection.

NIH Research Projects · FY 2026 · 2026-06

SUMMARY Disorders of the peripheral nervous system, particularly those that impair motor function or alter pain sensitivity, remain major clinical challenges. Current treatments are not curative and only manage secondary symptoms. Restoration of peripheral nerve function depends on the ability of Schwann cells to undergo transdifferentiation into repair cells—a complex cellular transformation that enables axonal regrowth and remyelination. This process requires tight coordination between extracellular signaling and transcriptional reprogramming. Despite advances in identifying molecular components involved in regeneration, therapeutic development has stagnated, in part because it remains unclear how injury-activated signaling cascades directly modulate transcription factor activity to drive this reprogramming. We are addressing this gap by focusing on Mitf, a member of the MiT/Tfe transcription factor family, which we have identified as an injury-induced regulator of Schwann cell plasticity. Our preliminary data show that Mitf is rapidly activated after nerve injury and is essential for repair cell formation, axonal regeneration, and sensorimotor recovery. Yet the mechanisms that govern its activation, localization, and transcriptional output remain unknown. With newly available genetic, imaging, transcriptomic, and proteomic tools, we can now dissect how injury-derived signals transmitted through the cytoplasm influence nuclear transcriptional programs to generate effective repair cells. We hypothesize that Mitf acts as a molecular integrator, linking injury-derived signals to gene expression programs that drive Schwann cell plasticity to promote peripheral neuron regeneration after injury; and that outside of acute injury it functions as a sentinel, continuously responding to nerve stress to preserve peripheral nerve integrity across lifespan. Here we propose to use integrated molecular and genetic approaches to dissect how Mitf governs Schwann cell plasticity and nerve regeneration. In this proposal we will: 1.) Identify the Mitf- dependent networks that drive Schwann cell transdifferentiation after injury 2.) Determine how phosphorylation status regulates Mitf nuclear localization and function. 3) Define the role of Mitf in maintaining Schwann cell homeostasis. These studies will establish how Mitf integrates injury-activated signaling cascades to effect transcriptional programs in peripheral glia, providing a mechanistic framework for Schwann cell plasticity and homeostasis.

NIH Research Projects · FY 2026 · 2026-06

Project Summary Sudden, life-threatening cardiac events can be terrifying experiences that can trigger the onset of posttraumatic stress disorder (PTSD). PTSD symptoms are common after cardiac events, with over 1 in 5 patients receiving implantable cardioverter defibrillators (ICDs) for prevention of life-threatening arrhythmias and sudden cardiac arrest exhibiting elevated PTSD symptoms. Cardiac disease-induced PTSD symptoms are associated not only with worse mental health but a worse clinical prognosis, including poorer quality of life (QoL), greater disability, and greater risk of event recurrence and/or all-cause mortality. Further, cardiac-focused anxiety sensitivity—a unique aspect of cardiac disease-induced PTSD—is linked to worse health, medical reassurance seeking, and greater healthcare utilization. However, evidence-based interventions for these cardiac-induced psychological presentations are lacking. This R34 proposal will draw from existing, evidence-based psychotherapies to develop a streamlined intervention that addresses cardiac trauma-related fear and cardiac-focused anxiety sensitivity— two key manifestations of PTSD after cardiac events that relate to adverse outcomes and are direct targets of gold standard, exposure-based interventions for PTSD and panic disorder. Trauma-focused exposure reduces trauma-related fear and interoceptive exposure reduces anxiety sensitivity, but neither intervention has been tested in patients with cardiac trauma. After initial intervention development (Stage Ia), a pilot randomized controlled trial (RCT) will be conducted (Stage Ib), in which patients with ICDs (N=70) from the UCLA Cardiac Arrhythmia Center will be randomized to 1) the trauma and anxiety sensitivity exposure-based treatment, called Cardiac Fears Treatment (CFT), or 2) standard supportive therapy (treatment as usual [TAU]). Participants will be assessed at pre-treatment, several periods throughout treatment, post-treatment, and a 6-month follow-up. In Aim 1, we will develop the CFT intervention using an iterative approach, guided by prior research and qualitative interviews with key stakeholders (e.g., cardiology experts), and pilot testing in a small (N=8) open trial of patients with ICDs will yield initial feasibility and acceptability data. Subsequent aims will analyze data from the pilot RCT, generating additional feasibility and acceptability data, along with preliminary efficacy information. In Aim 2, we will compare CFT to TAU on psychological responses (self-report and behavioral task measures of cardiac trauma-related fear and cardiac-focused anxiety sensitivity). Aim 3 will compare CFT to TAU on health- related outcomes (health-related QoL and healthcare utilization). Finally, in Aim 4, we will explore whether key psychological processes targeted in the intervention (cardiac trauma-related fear and cardiac-focused anxiety sensitivity) mediate change in health-related outcomes. By targeting key psychological processes associated with adverse outcomes after a cardiac trauma, this mechanism-focused study has the potential to improve the emotional and physical health of cardiac patients and will generate critical feasibility, acceptability, and preliminary efficacy data needed to inform a grant proposal to test a refined version of CFT in a larger RCT.

NIH Research Projects · FY 2026 · 2026-06

Project Summary Despite decades of research effort, the HIV-1 retrovirus remains a major global public health crisis. To date, there is no cure for HIV-1, with an infected individual requiring continuous antiretroviral therapy (ART) to thwart the onset of AIDS. When ART is ceased, there is a rebound of viremia that originates from the pool of latently infected CD4+ T cells. This rebound within a latently infected cell begins with transcriptional reactivation of the HIV-1 genome, and the Transactivator of transcription (Tat) protein lies at the heart of HIV-1 gene expression. Tat activates gene expression by recruiting the kinase Positive Transcription Elongation Factor b (P-TEFb) to a paused RNA Polymerase II (RNAP II), ultimately promoting RNAP II pause release. Specifically, Tat recruits P- TEFb to the nascent Transactivation Response (TAR) RNA element, with Tat positioning P-TEFb for phosphorylation of the RNAP II C-terminal domain and transcription factors to promote RNAP II pause escape. Importantly, a major cellular reservoir of P-TEFb is held inactive by the dimeric Hexim1 (or less commonly Hexim2) protein within the 7SK ribonucleoprotein (RNP), and prior work supports that Tat “hijacks” P-TEFb from 7SK RNP for subsequent P-TEFb recruitment to TAR. However, the molecular mechanisms underpinning this hijacking remain uncharacterized, posing a fundamental barrier to targeting Tat transactivation as an HIV-1 eradication strategy. A potential route to Tat removal of P-TEFb is by competing with Hexim1 for 7SK RNA binding, since one of the Hexim1 binding sites on 7SK RNA mimics the Tat recognition site on TAR RNA. Therefore, Aim 1 proposes to define the competition between Tat and Hexim1 for binding to 7SK RNA in vitro by NMR, ITC, and EMSA experiments, building upon recent work from the Feigon lab that successfully employed these same methods to identify Hexim1 binding sites on and affinities for 7SK RNA. Aim 2 proposes to purify and biochemically characterize a Tat-bound 7SK RNP from mammalian cells, since this complex is a proposed intermediate prior to Tat–P-TEFb handover to TAR. Specifically, the complex will be characterized with mass photometry, mass spectrometry, and SHAPE-/DMS-MaP. Lastly, Aim 3 describes how the structure of a Tat– 7SK RNP will be determined by cryoEM and how structure-derived hypotheses will be tested on Tat transactivation in a cell line model of HIV-1 latency. The proposed work is supported by the Feigon lab’s excellent track record of combining NMR and cryoEM to study challenging, dynamic RNPs and, all the necessary facilities and equipment are available to execute this work. Together, the goal of this fellowship proposal is to uncover the structural and mechanistic basis of Tat “hijacking” of P-TEFb from the 7SK RNP, providing new insights to aid design of HIV-1 therapeutics and offering a rich training in structural biology methods.

NIH Research Projects · FY 2026 · 2026-06

SUMMARY/ABSTRACT Cranial radiotherapy remains a cornerstone for treating primary and metastatic brain tumors. However, it is frequently associated with long-term cognitive decline, particularly in patients who survive beyond six months post-treatment. Despite advances in radiation delivery and neuroprotective strategies, no effective interventions currently exist to prevent or reverse radiation-induced neurotoxicity. The studies outlined in this proposal are based on a novel hypothesis that radiation disrupts neural lineage commitment in the developing human brain, driving aberrant cellular plasticity and transdifferentiation of neural stem/progenitor cells into endothelial-like phenotypes, which is supported by preliminary and published data. These fate shifts impair neurodevelopment and alter synaptic function, contributing to long-term cognitive deficits. The overall hypothesis is that radiation-induced cellular reprogramming in the human brain alters both cellular identity and functional connectivity, and that these effects can be mitigated through targeted inhibition of endothelial-inductive pathways (e.g., VEGF, FGFR, Notch). To test this, we will use mature human iPSC-derived cortical organoids subjected to fractionated radiation and apply single-cell and spatial transcriptomic approaches to define lineage transitions (Aim 1), followed by functional assays including calcium imaging and multielectrode array recordings to assess neural network activity in the presence and absence of pathway inhibition (Aim 2). This R21 project leverages cutting-edge 3D human brain models and state-of-the-art technologies to address a major gap in understanding the cellular mechanisms underlying radiation-induced cognitive impairment. Because the targeted pathways have existing pharmacologic inhibitors already in clinical use, this research has high translational potential for developing radiation mitigators to preserve brain function in cancer patients undergoing radiotherapy.

NIH Research Projects · FY 2026 · 2026-06

PROJECT SUMMARY/ABSTRACT 80% of those affected by autoimmune diseases are female. Yet, the biological basis for female predisposition to autoimmunity remains unclear. Recent work has uncovered a role for Xist, a female-specific, long non-coding RNA, in activating B cell autoantibody production and stimulating innate immune cells. However, it remains unclear the role of Xist in promoting T cell pathogenicity. This is a significant gap in knowledge, since CD4+ T cells play a critical in multiple female-biased autoimmune conditions. including Crohn’s Disease. In a mouse model of autoimmune colitis that shares multiple features with Crohn’s, we demonstrate that T cell-specific Xist deficiency protected against colitis induction and was associated with lower CD4+ T cell numbers and decreased IFN-g production. Consistent with Xist’s well-known role in X chromosome inactivation, Xist-deficient CD4+ T cells exhibited upregulation of a number of X-linked immunoregulatory genes. Moreover, multiple immunoregulatory genes on autosomes were upregulated in Xist-deficient CD4+ T cells, suggesting that Xist may spread beyond the X chromosome to regulate autosomal genes. Thus, we hypothesize that Xist promotes autoimmune disease development by increasing CD4+ T cell numbers and IFN-production through X-linked and autosomal gene regulation. In Aim 1, we will define the molecular mechanisms by which Xist controls X-linked and autosomal gene transcription in mouse CD4+ T cells. In Aim 2, we will determine how Xist alters CD4+ T cell numbers and cytokine production during colitis development in a mouse model of Crohn’s Disease. In Aim 3, Xist’s role in regulating the transcriptional profile, apoptosis propensity, and cytokine production in human CD4+ T cells from female patients with Crohn’s disease will be determined. Knowledge gained from completion of these studies will contribute to our basic understanding of Xist biology and mechanisms by which female- specific Xist controls T cell autoimmune pathogenicity. A deeper understanding of factors promoting autoimmunity in females will benefit human health overall by revealing new targets for immunotherapy and guiding interventions for optimizing tolerance induction.

NSF Awards · FY 2026 · 2026-06

The sense of hearing shows a truly remarkable ability to detect weak sounds amidst loud and competing environmental noise. This detection is enabled by hair cells, specialized cells that are responsive to mechanical movement below that of a nanometer. Prior work has shown that these cells are active – they expend energy to enhance their response. Secondly, they are not linear – they amplify weak signals and compress strong ones. Other studies have uncovered various cellular processes that contribute to the hair cell’s ability to sense a very weak signal. However, what is not known is how these very weak signals are differentiated from various noise sources, both within the biological system itself and the external environment. The investigators plan to measure the hair cell response to complex sounds, starting from pure tones and increasing complexity to mimic sounds that the animals would be detecting in nature. Information theory and other statistical methods will then be used to determine which classes of signals the cell is most responsive to. The investigators will also test how this response is degraded by the presence of noise. As a comparison, they will use hair cells both from an auditory organ, specialized for detecting animal calls and other air-borne signals, and from a vestibular organ, specialized for detecting ground-borne signals. This study explores how auditory and vestibular systems perform remarkably sensitive detection in the presence of noise and competing background signals. Analytical methods from the field of information theory will be adapted to study the graded response exhibited by mechanosensitive cells, with measurements obtained from sensory epithelia in vitro. Specifically, transfer entropy, Kolmogorov entropy, and mutual information will be applied to determine the degree of information extracted by sensory cells of the auditory and vestibular epithelia. The investigators will use semi-intact biological preparations of the inner ear, designed to maintain live and functional hair cells. External input of varying complexity, ranging from pure tones to transient pulses, as well as signals that mimic biologically relevant sound streams, will be applied to the cells. The goal of the study will be to determine whether there are classes of stimuli that optimize the information transfer. The robustness of the cellular response to noise and competing streams of information will be directly assessed by superposing these interfering signals onto the test stimuli and measuring the rate at which the information transfer is degraded. Computational models of the sensory cells will then be utilized to explain how the cells expend energy to perform computation, thus enabling the extraction of weak signals from the surrounding noise. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2026 · 2026-06

Multimodal datasets, which combine sources such as medical imaging, clinical records, and genetic information, have the potential to significantly advance our understanding of complex systems and improve health outcomes. However, the heterogeneity, high dimensionality, and lack of reliable statistical tools often lead to unstable analyses or misleading conclusions. These issues — and the limited ability to rigorously quantify uncertainty or disentangle relationships among data sources — pose a major barrier to the adoption of data-driven methods in high-stakes settings, where the cost of error can be substantial (e.g., clinical decision-making, disease monitoring, and health policy). This research project will develop robust, scalable, and statistically principled methods for integrating and analyzing multimodal data, with a particular emphasis on uncertainty quantification. The project also integrates research and education through: (a) the involvement of undergraduate, graduate, and postdoctoral students in both research and dissemination, along with mentoring to support their continued professional development; (b) the integration of research findings in UCLA courses and openly accessible online materials; and (c) workshops and outreach activities designed to broaden participation in data science. In more detail, the research focuses on the challenge of nonparametric estimation and uncertainty quantification for multimodal data, in which multiple high-dimensional and heterogeneous data sources must be integrated to enable reliable inference. The initial goal is to develop robust and scalable methods for estimating the effects of individual modalities, utilizing deep learning to model auxiliary structures and employing kernel-based techniques to provide uncertainty quantification. Armed with such methods, the follow-up goal is to construct machine learning-powered estimators that identify and quantify the pathways through which modalities influence outcomes, combining multi-level Monte Carlo techniques with deep learning predictions and observational data to uncover complex mediator effects. A third goal is to extend these approaches to longitudinal settings, enabling inference in the presence of temporal dynamics and high-dimensional confounding. The work in this project will leverage and build upon techniques and tools from deep learning, empirical process theory, Monte Carlo methods, and reproducing kernel Hilbert spaces to ensure both statistical rigor and computational efficiency. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2026 · 2026-06

PROJECT SUMMARY Mechanical forces are crucial regulators of diverse biological processes, such as sensory perception (e.g., hearing, balance, touch), blood pressure regulation, and essential cellular behaviors like migration, division, and differentiation. As a pivotal biophysical parameter, membrane tension critically governs cell behavior by dynamically modulating ion channel activity, cytoskeletal organization, and mechanotransduction pathways. Current tools for measuring dynamic membrane tension changes in living cells, especially in rapid sensory systems like the cochlea, are limited in speed, sensitivity, and specificity. This project addresses this critical gap by developing PresTen, a novel, cutting-edge, genetically encoded fluorescent biosensor. PresTen is engineered from prestin (SLC26A5), a unique motor protein in cochlear outer hair cells responsible for rapid electromotility. Prestin’s inherent mechanosensitive properties and conformational changes make it exceptionally suited for detecting dynamic membrane tension changes without perturbing ion gradients. Our recent finding that prestin responds to membrane tension and thinning provides an intrinsic advantage, offering superior sensitivity and speed over conventional reporters. This exploratory tool-development project aims to establish PresTen’s feasibility and utility. We will engineer and optimize PresTen for enhanced performance by uncoupling prestin’s voltage/chloride sensitivities and integrating circularly permuted fluorescent reporters into mechanosensitive domains. Subsequently, we will calibrate PresTen’s fluorescence response against defined membrane tension using reconstituted systems and live-cell measurements. Finally, we will validate PresTen in key mechanotransduction systems, applying it to cochlear outer hair cells to capture high-frequency force dynamics and in cardiomyocytes to demonstrate broader applicability. Upon its successful development, PresTen will revolutionize our ability to precisely measure dynamic membrane tension in live cells and tissues, offering an unprecedented high-speed, non-invasive platform. This novel bioengineered tool will transform our understanding of mechanotransduction, opening new avenues for discovery in diverse areas from sound amplification and cardiac disease to immune cell activation and cancer cell metastasis. Crucially, PresTen also creates novel technological opportunities, serving as a precise mechanical force indicator for nanodrug delivery systems and for high-throughput drug screening, where direct membrane tension monitoring is currently a major challenge. This R21 project represents a critical, innovative step towards unraveling the dynamic roles of mechanical forces in health and disease.

NIH Research Projects · FY 2026 · 2026-06

PROJECT SUMMARY Suicide is the third leading cause of death for youth aged 10 to 28, with Latine adolescents and young adults experiencing disproportionately high rates of suicidal thoughts and behaviors (STBs). Despite this, Latine youth are significantly underrepresented in suicide research. This R21 project aims to identify key risk and protective factors influencing the developmental trajectories of STBs in an existing sample of 674 Mexican-origin youth, followed longitudinally from age 10 to 28. The proposed research leverages data from the California Families Project (CFP), a 19-year longitudinal study that provides a robust, multi-informant, and multi-method dataset. This study is innovative in its focus on the interaction between clinical symptoms and sociocultural factors, and their combined influence on the trajectories of STBs across critical developmental stages. It is one of the first studies to comprehensively explore STB trajectories in Mexican-origin youth, a high-risk yet understudied group. Guided by the Family Stress Model (FSM), a culturally informed framework, this study will address significant gaps in the literature by examining STB trajectories, clinical predictors, and culturally relevant protective factors. Aim 1 will chart the longitudinal trajectories of STBs from preadolescence through young adulthood. Aim 2 will investigate how key clinical symptoms (depression, anxiety, substance use) predict the initial levels and rate of change in STBs. Aim 3 will identify sociocultural factors (ethnic pride, familismo) that compensate for or moderate the impact of clinical symptoms on STB trajectories. This research is significant because it has the potential to inform the development of culturally responsive and developmentally informed suicide interventions for Latine youth. By understanding the sociocultural context and specific risk pathways in Latine families, the findings could advance scientific knowledge and lead to targeted, effective strategies to reduce suicide rates among high-risk youth populations. The study’s innovative approach and the use of a rich longitudinal dataset underscore its feasibility and relevance to the NIMH mission of improving mental health outcomes and reducing suicide disparities for Latine youth.

NIH Research Projects · FY 2026 · 2026-06

PROJECT SUMMARY/ABSTRACT This proposal presents a research program focused on the utilization of pH imaging, surgery, and pH modulation for the treatment of infiltrative glioblastoma (GBM). Despite therapeutic advances in oncology, isocitrate dehydrogenase wild-type (IDHwt) GBM remains resistant to current treatment modalities, with median survival rarely exceeding 24 months post-diagnosis. Dysregulation of pH homeostasis, a hallmark of malignant progression extensively characterized in extracranial neoplasms, represents a promising target for both diagnostic imaging and therapeutic intervention. We have developed and validated amine chemical exchange saturation transfer echo-planar imaging (CEST-EPI), a pH-sensitive magnetic resonance imaging technique that enables high-resolution visualization of metabolically active tumor regions through detection of extracellular acidosis in both contrast-enhancing and non-enhancing infiltrative zones of GBM. Preliminary data demonstrate CEST-EPI's enhanced sensitivity and specificity for detecting infiltrative tumor regions compared to conventional imaging modalities. This technique has enabled characterization of differential expression patterns of key pH regulatory mechanisms, including monocarboxylate transporters (MCT1/4), sodium-hydrogen exchangers (NHEs), and sodium-bicarbonate cotransporters (NBCs) across varying pH microenvironments. Based on these findings, we hypothesize that CEST-EPI-guided surgical resection targeting infiltrative tumor regions beyond conventional margins is both safe and feasible, while simultaneously providing a platform for investigating pH regulatory mechanisms in GBM. Furthermore, we propose that targeted inhibition of these mechanisms represents a novel therapeutic strategy. Our multidisciplinary research program integrates expertise across advanced neuroimaging, neurosurgical techniques, and preclinical model development to establish a novel therapeutic paradigm for targeting pH dysregulation in GBM. Successful validation of this approach could significantly improve local tumor control with potential implications for enhanced overall survival, representing a paradigm shift in the therapeutic landscape of GBM management. The combination of innovative imaging technology, molecular validation approaches, trial design, and preclinical models positions this work to significantly advance our understanding of pH regulation in glioblastoma while developing new therapeutic strategies. Success could establish a new paradigm for metabolic imaging-guided surgery and pH-targeted therapy in brain tumors.

NIH Research Projects · FY 2026 · 2026-06

Project Summary/Abstract The incorporation of difluoromethyl (–CF2H) group has become a leading strategy in pharmaceutical development to enhance drug performance due to –CF2H’s unique and profound bioisosteric activity for hydroxyls, amines, and thiols as a result of its hydrogen bond ability. Presently, transition metal-mediated approaches for the direct incorporation of the –CF2H group is an attractive methodology in the context of selective late-stage functionalization of complex molecules. While there is an abundance of synthetic methods employing this approach for the formation of the traditional carbon(sp2)-CF2H bonds, general protocols for the formation of more challenging carbon(sp3)-CF2H bonds are still limited. Hence, there is an urgent and unmet need to develop new, sustainable, and efficient synthesis methods to accelerate the synthesis and evaluation of bioactive compounds containing C(sp3)-CF2H bonds. The hallmark of this proposal is the development of three-component iron-catalyzed difluoromethylation reactions mediated through iron metal for the formation of C(sp3)–CF2H bonds. To achieve this objective, I will put forward two strategies. The strategies differ not in the specific C(sp3)–CF2H bond that is to be formed, but in the distinct, yet complementary, mechanistic manifold of introducing the difluoromethyl group into the organic substrate. The first approach employs HCF2–X electrophiles as •CF2H radical precursors which can, through strategic manipulation of reaction conditions, engage with a “radical lynchpin” and subsequently C(sp3)–CF2H bond formation with a well-defined iron-species (Aim 1). The second strategy will implement the use of organozinc reagent, [(DMPU)2Zn(CF2H)2], as a nucleophilic source of –CF2H group to be transmetalated into the iron catalytic cycle as a complementary route to synthesize the C(sp3)–CF2H bond (Aim 2). To bolster these efforts, I plan to implement advanced spectroscopic techniques (Mössbauer spectroscopy and single-crystal X- ray diffraction) as well as computational chemistry to accelerate the development of three-component iron- catalyzed difluoromethylation reactions. While both approaches target the development of iron-mediated synthetic methods that permit the direct incorporation of the –CF2H group, the mode of introduction is different. In doing so, this research project will foster the growth in knowledge in iron catalysis. The long-term goal of this proposal is to expand the fundamental research of domestically manufactured metals in the U.S., such as earth-abundant iron, for the development of synthetic methods that permit expedient access to high-value molecules in an economical, sustainable, and safe manner all while fostering and refining our knowledge and understanding of it.

NSF Awards · FY 2026 · 2026-06

This project aims to develop theory and methodology required for statistical inference on spatiotemporal rates of change or gradients, followed by extending their use to assess boundaries that track significant changes in spatiotemporal response. The current stage of spatial and temporal data science bears witness to the recording of massive spatiotemporally indexed data for the purpose of tracking changes in spatial and temporal variables. This project outlines the details of the methodology and software development for quantifying and understanding change within large and complex spatiotemporally referenced datasets. These are closely related to machine learning and artificial intelligence, and the developments are motivated by substantive questions arising in various fields where assessing regions of rapid change in space and time is crucial. The focus of applications is on biomedical and neuroimaging datasets, and the project provides research training opportunities for graduate students. Extending the statistical inference to larger domains, we leverage a low-rank projection-based approximation to exact Gaussian processes. The project will also develop classes of highly scalable Bayesian factor models and Graphical predictive processes for jointly modeling highly multivariate spatiotemporal data. The project will conduct rigorous investigations into statistical inference for rates of change associated with predictive processes and graphical predictive processes. The project aims to derive probability distributions that facilitate posterior inference within a Bayesian setting. This is followed by extending the inference to smooth surfaces within space-time that track rapid directional change. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2026 · 2026-06

Summary Genetic ataxias are slowly progressing neurodegenerative diseases causing severe disability for which no disease-modifying therapy exists. Though individually they are rare diseases, together they affect nearly 15,000 people worldwide. Current clinical scales for assessment of ataxia severity and progression are ineffective for evaluation of treatment effects during the timeframe of typical clinical trials, necessitating inclusion of high numbers of patients, which often is not feasible, to achieve statistically meaningful outcomes. These challenges could be addressed if reliable, sensitive, and preferably minimally invasive biomarkers for ataxia severity and progression were available for use in clinical trials. Unfortunately, such biomarkers do not presently exist. Therefore, we propose an initial study combining both biomarker discovery using proteomic and transcriptomic approaches, and testing of several promising candidate biomarkers in blood samples from patients with the most common dominant and recessive types of genetically inherited ataxia. We will utilize state-of-the-art techniques including Olink®’s proximity extension proteomics, NextGen RNA sequencing, and electrochemiluminescence immunoassays. The analyses will be done in patient plasma and neuronal extracellular vesicles isolated from the plasma, increasing the likelihood of discovering biomarkers reporting on specific biochemical changes in the central nervous system. The study will generate an initial set of potential biomarkers, providing the basis for subsequent, larger testing and validation in the context of R01 or U01 applications, addressing a current urgent gap in developing effective therapies for patients with genetic ataxias. Additionally, the proteomic and transcriptomic data will allow pathway analysis that may shed new light on the mechanisms underlying the pathogenesis of specific types of genetic ataxia, including both common and distinct features among them.

NIH Research Projects · FY 2026 · 2026-06

Project Summary Hashimoto’s Thyroiditis (HT) is a prevalent autoimmune disease affecting 15% of the population and characterized by chronic autoimmune attack on thyroid follicular cells. Over time, this persistent autoimmune attack results in thyroid gland failure and the requirement for lifelong hormone replacement, a hallmark shared with other chronic autoimmune diseases such as Type 1 Diabetes Mellitus (T1DM) and Addison’s Disease. Despite its prevalence, the mechanisms driving the unrelenting autoimmune attack in HT remain poorly understood. Identifying factors underlying the persistent autoimmune response, may identify new therapeutic targets to halt disease progression in HT and similar chronic autoimmune disorders. We have identified a population of TCF7+ stem-like progenitor CD8 T cells within the thyroid tissue of HT patients, analogous to those seen in chronic viral infections and Type 1 Diabetes, that sustain autoimmune attack by replenishing the pool of terminally differentiated cytotoxic effectors. Using single cell RNAseq and TCRseq of human thyroid specimens from individuals with HT, our preliminary data further demonstrate the transcriptional transition and clonal expansion of TCF7+ progenitor CD8 T cell to effectors with killing ability within the thyroid. In addition, our preliminary data suggest that tertiary lymphoid structures (TLSs), organized collection of immune cells within the thyroid, provide a microenvironment that promotes autoimmunity, driven in part by CD4 T follicular helper (Tfh) cells and IL-21, a cytokine implicated in CD8 T cell differentiation. Using a mouse model of HT, we demonstrate that IL-21R deletion protects against thyroid autoimmunity, suggesting that TLS-associated factors may drive the conversion of progenitor CD8 T cells into cytotoxic effectors. Thus, our overarching hypothesis is that TCF7+ CD8 T progenitor cells sustain autoimmune persistence in HT, while TLS- associated signals promote their differentiation into cytotoxic effectors, perpetuating disease progression. In Aim 1, we will define the role of TCF7 in maintaining stem-like CD8 T cells by genetically deleting TCF7 in an HT mouse model and using WNT pathway agonists to assess its regulatory function. In Aim 2, we will investigate TLS-driven CD8 T cell conversion by assessing the deletion of CD4 Tfh cells and IL- 21 signaling in HT progression using mouse models. We will leverage spatial transcriptomics data previously collected by our lab from HT thyroid specimens, to determine how TLS localization influences CD8 T cell differentiation. These studies will uncover fundamental mechanisms of chronic autoimmunity in HT, providing potential therapeutic targets to disrupt persistent autoimmune attack and yielding broader insights into T cell– mediated autoimmune diseases.

NIH Research Projects · FY 2026 · 2026-05

PROJECT SUMMARY Radioembolization (RE) with yttrium-90 (90Y) microspheres is a promising treatment for primary and secondary liver cancer, but the success of the treatment is user-dependent and relies on the optimal distribution of 90Y particles within the tumor. The current method for determining how 90Y microspheres are dispersed in a tumor requires the tracking of Tc-99m-labeled macro-aggregated albumin (MAA) distribution after the mapping procedure. This methodology is labor-intensive and cumbersome, and does not account for flow dynamics within the tumor. Variations in peri-tumoral and intra-tumoral blood flow can significantly impact final 90Y microsphere distribution, leading to undertreated cold spots or overtreated embolic regions that cause beads to reflux backwards. There is thus an unmet need for an advanced tool to predict 90Y distribution accurately based on tumor flow characteristics. This project aims to overcome this current barrier by developing an in vivo imaging biomarker of 90Y microsphere distribution that leverages AI-optimized perfusion analyses and a biological model of liver flow dynamics. The project will be conducted via two aims: 1) Development of an AI-optimized liver perfusion analysis tool using routine clinical MRI scans. Quantitative dynamic contrast-enhanced (DCE) MRI will be used to obtain insights into tumor blood flow and vascularity. With standardized data collection and curation, we will develop advanced deep learning-enabled perfusion analysis to analyze liver perfusion characteristics in routine clinical DCE-MRI. A physical model of particle flow will be integrated with perfusion MRI for accurate prediction of RE microsphere distribution, and preclinical evaluation will be conducted with patient- derived liver tumor phantoms, allowing for precision measurements of the MRI-based perfusion protocol. 2) In vivo validation of AI-optimized perfusion protocol for accurate prediction of 90Y particle densities. We will test intra-subject and inter-subject variability of the perfusion MRI protocol by repeating DCE-MRI scans at multiple time points with concomitant repositioning. Once intra-subject and inter-subject reproducibility are validated, the AI-optimized perfusion MRI will be leveraged to quantify perfusion parameters in an in vivo porcine liver cancer model intra-arterially infused with 90Y microspheres. Histopathological and imaging analyses of 90Y microsphere distribution will then be correlated with the MRI perfusion markers against those from the standard Tc-99m MAA administration. The results of this project will have significant implications for the treatment of liver tumors by providing a foundation for more effective and precise 90Y-based therapies. By optimizing 90Y particle density with tumor perfusion characteristics, we can improve the outcomes and safety of RE, ultimately transforming it into a more effective and precise therapeutic approach for liver cancer patients.