Florida State University

NSF Awards · FY 2026 · 2026-10

Many modern Artificial Intelligence (AI) models can produce meaningful text, but they often fail on complex structural and numerical data involving different units, formulas, information describing the data (i.e., metadata), and hierarchies. These failures are especially concerning in areas such as medicine, finance, defense, and space, where even small quantitative mistakes can lead to misleading conclusions and significant negative consequences. This project aims to address this problem by developing a new AI model focused on accurate comprehension of complex numerical and structured data rather than natural language text. The project will help make scientific knowledge more transparent and accessible, while also supporting education through new teaching materials, student research opportunities, and outreach activities that engage learners in data reasoning. By improving the ability of AI to work correctly with complex numerical and structured data, the project advances the progress of science, supports health and welfare, and strengthens the nation’s capacity for trustworthy data-driven discovery and decision-making. The project develops the Large Number Model (LNM), a hybrid neural-symbolic model for reliable reasoning over numbers, units, formulas, and complex tabular data. The research includes three main activities: creating scalable methods to extract numerical and structured information from documents, designing model architectures that represent quantities and two-dimensional tabular structures more effectively than text-only systems, and incorporating symbolic validation to check algebraic, dimensional, and semantic consistency. The project will also develop methods for combining quantitative evidence across multiple sources and will evaluate the resulting system through controlled experiments, robustness tests, and benchmark datasets drawn from scientific and medical domains. The expected contribution is a new foundation for AI systems that are more accurate, interpretable, dependable, and compatible with the full data cycle when working with complex numerical and structured knowledge. This, in turn, is expected to maximize the utility of information resources. 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

Genetically identical individuals, such as human twins, are expected to be very similar, but they nonetheless exhibit substantial differences in traits and diseases throughout their lives. In controlled laboratory settings where all individuals experience the same environment, genetically identical animals also exhibit differences in important traits, including fertility and lifespan. However, why genetically identical individuals differ from each other for these traits remains incompletely understood. In addition, how non-genetic differences that arise in one generation impact subsequent generations is not well understood. This research builds upon the principal investigator’s prior work, which revealed that differences in how genes are regulated are linked to differences in fertility across genetically identical individuals and that non-genetic differences in one generation can lead to predictable differences in subsequent generations. Using a powerful set of experiments, this project will investigate how DNA and other factors interact to control differences among genetically identical individuals. This research priority will advance scientific progress in biotechnology and artificial intelligence, as well as contribute to national health, because many diseases are not just regulated by the DNA sequence alone, but by interactions with other factors. Genetically identical, i.e. isogenic, individuals exposed to the same environment can develop at different rates, grow to different sizes, and produce differing numbers of progeny. Such phenotypic variability raises key questions: 1) what drives variability across isogenic individuals in the same environment?; 2) how do populations differentially evolve when they are isogenic but epigenetically distinct?; and 3) how can phenotypes of isogenic individuals be better predicted? This research will use a highly integrated set of three objectives to answer each of these questions, all using isogenic populations within the roundworm Caenorhabditis genus, including the widely used genetic model, C. elegans. The project will specifically identify conserved genomic and epigenomic features that control gene expression variation in isogenic populations, elucidate how isogenic individuals with distinct epigenetic states evolve over hundreds of generations, and improve phenotypic predictions using reproductive and transcriptomic data. This project will integrate high-throughput sequencing data, causal analysis of gene function (RNA interference), and advanced analytic tools (artificial intelligence/machine learning) to gain insights on these objectives. Integrated with these research objectives, this project will implement an educational plan to provide meaningful research experiences, individual mentoring, and bioinformatics training to high school students, undergraduate students, graduate students, and postdoctoral researchers, contributing to the advancement of the STEM workforce of the United States. Collectively, the questions posed are of high interest for fundamental understanding how an individual develops its traits from a combination of its genome, epigenome, and transcriptome. The proposed objectives are carefully designed to address this using cutting-edge high-throughput approaches and novel analytical methods while encouraging the development of talent in STEM fields. 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

Different species often arrive at similar solutions to recurring challenges through convergent evolution. Such events provide some of the strongest examples of adaptation, yet understanding the molecular mechanisms and constraints that result in such convergence has only recently become possible due to advances in computing and genomics. Among animals, traits such as flight have arisen independently numerous times (e.g., in lineages giving rise to birds, bats, and insects), defining the ecologies and enabling the success of the resulting species. Similarly, venoms have arisen independently more than 100 times in animals and play diverse roles in, for example, predation and defense. These recurring traits represent optimal systems for investigating the rules and limitations of how evolution can yield complex adaptations, a major open challenge in evolutionary biology. This project will use integrative approaches including AI in multiple venomous animals to understand how complex traits repeatedly arise and evolve. Such an approach will enable not only the identification of how complexity originates but also catalyze future biotechnological innovations in the bioeconomy by uncovering functional solutions to common problems across the Tree of Life. Convergent evolution is a hallmark of adaptation and provides a means for delineating the roles of genetic and functional constraints in determining evolutionary trajectories. Venoms are one of the most common and convergent functions among animals, with more than 200,000 venomous species from more than 100 venom-origin events, and venom function requires recurrent evolution of specialized tissues and gene-regulatory networks to express, process, secrete, and store toxins. Substantial convergence in recruited protein families, tissues of origin, and contributing gene-regulatory networks has been observed, yet venoms are exceptionally variable at all taxonomic levels. Venoms therefore represent a unique opportunity for discerning rules and idiosyncrasies of complex trait origin and subsequent evolution under parallel constraints. Eighteen species representing three independent venom origins in centipedes, scorpions, and snakes will be used to investigate the impacts of deep evolutionary events during trait origins on ongoing complex trait evolution. A hierarchical phylogenetic framework will be used to link macro- and microevolutionary processes, focusing on how deep-origin events bias evolutionary trajectories. Overall, our sampling strategy will allow us to bridge macro- and microevolutionary processes and investigate convergence at multiple biological (genome, tissue, organismal) and phylogenetic (within species, across species, and across lineages) scales, specifically focusing on how trait origin and secondary innovation events influence ongoing evolution among close related species. 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

This project addresses a fundamental challenge in understanding how genetics influences human health and behavior: distinguishing true biological genetic effects from influences that arise through family environment. Human genetic studies often ignore environmental factors, yet growing evidence suggests that parental characteristics can shape the family environment and confound genetic associations. As a result, common analytic approaches may overstate genetic contributions or misidentify biological mechanisms. This project develops new statistical tools to separate direct genetic effects from environmentally mediated influences, enabling more accurate interpretation of genetic association findings. The results will improve the reliability of genetic risk prediction through biotechnology, inform precision medicine, and support evidence-based public health and policy decisions. The project will also produce open-source software and provide interdisciplinary training opportunities at the interface of statistics, biostatistics, genetics, and data science. The research develops a unified statistical framework grounded in causal inference and high-dimensional data analysis. First, it introduces a new definition of heritability based on counterfactual comparisons between individuals with identical environments, allowing direct genetic contributions to be isolated and bounded under realistic assumptions. Second, it develops methods to estimate direct genetic effects by combining large population-based studies with smaller family-based studies using summary-level data. This approach leverages the strengths of both study designs to produce unbiased and statistically efficient estimates without requiring extensive family data. The methods will be supported by theoretical guarantees, scalable algorithms, and applications to large genomic datasets. By providing principled tools for disentangling genetic and environmental effects, the project advances both statistical methodology and the scientific understanding of complex traits. 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

Collaborative inference at the network edge enables low-latency and privacy-sensitive artificial intelligence (AI) services without relying on remote cloud infrastructure. Distributing increasingly complex neural network models across nearby edge devices enables the pooling of their compute and memory resources to perform inference beyond the capability of any single device. Existing collaborative inference approaches rely on centralized or static control under assumptions of stable network connectivity, leading to inefficient resource use and degraded inference performance when network or system conditions change. This project reconceptualizes the wireless network not merely as a communication medium, but as a coordination substrate to orchestrate model execution and resource allocation for efficient, scalable, and resilient collaborative inference. The resulting advances will support emerging applications such as mobile health, distributed robotics, and intelligent transportation systems. The project also integrates research into teaching through new courses with hands-on learning modules on edge intelligence and distributed AI. It further strengthens an existing mentoring pipeline spanning K-12 outreach, undergraduate research participation, and graduate training, preparing students for future careers across AI, systems, and networking. The project advances the scientific foundations of network-aware collaborative inference at the intersection of networking, distributed systems, and AI. Low-cost decentralized awareness of network and device conditions, maintained and shared by edge devices, serves as a unifying foundation across the three research thrusts. Thrust 1 develops a decentralized inference framework that jointly optimizes model partitioning and resource allocation under partial observability for efficient and scalable general-purpose inference. The growing demand for generative inference with billions of parameters and autoregressive decoding introduces new system challenges. Thrust 2 therefore develops pipeline parallelism strategies for coordinating pipeline execution and inter-edge communication to improve the efficiency of collaborative inference under limited bandwidth and constrained edge resources. Thrust 3 develops model-agnostic runtime coordination mechanisms that sustain both general-purpose and generative inference performance under connectivity disruptions and device failures. The research combines algorithm design, systems prototyping, and experimental evaluation in realistic edge environments. Together, these advances will help shape future intelligent networked systems that integrate communication, computing, and AI. 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

Mountains are among the most important regions on Earth for both people and nature. They supply freshwater to cities and towns far beyond the mountains themselves, influence regional climate, and support nearly half of the world’s biodiversity hotspots while also sustaining about 12% of the global population. Mountain biodiversity also supports essential services such as food resources, medicinal discoveries, and livelihoods. Yet scientists still do not understand why some mountain regions host exceptional biological diversity, whereas others with similar environments do not. Without this knowledge, it is difficult to identify which areas are most valuable to protect, prioritize conservation investments, or anticipate how mountain ecosystems may change over time. This project will clarify how the physical formation of mountains over geologic timescales interacts with the behavior and adaptation of species to produce and maintain biodiversity. Results will provide key knowledge for long-term environmental planning and management of natural lands. This collaborative research will train undergraduate and graduate students, as well as a postdoctoral fellow, fostering a new generation of scientists fluent in evolutionary biology, bioinformatics, and geosciences. It will also create educational resources for K-12 students, develop hands-on learning experiences for families and local communities, and promote interdisciplinary collaboration through workshops and conference symposia that connect Earth and Life Sciences. This interdisciplinary project will investigate how geological dynamics and biological interactions jointly shape biodiversity by integrating newly assembled trait datasets with phylogenetic comparative methods and geospatial analyses. The study focuses on mountain squamates (lizards and snakes), a diverse group of over 12,000 species whose physiology and ecology are closely tied to environmental conditions. First, the project will quantify global patterns of taxonomic, morphological, physiological, and ecological diversity across mountain regions to determine how species differ in form, function, and niche occupation. It will then test how tectonic history and climatic variability influence lineage diversification within and among mountain systems. Finally, the project will evaluate how landscape dynamics interact with interspecific competition to shape phenotypic evolution. Together, these analyses will provide a more complete and mechanistic understanding of the processes that promote, constrain, or erode biodiversity in mountains and establish a framework broadly applicable to other organisms and dynamic landscapes worldwide. 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

NONTECHNICAL SUMMARY Imagine materials that look like tiny, patterned quilts at the atomic scale; these are called moiré materials. Their unique structure creates energy conditions (called flat bands) where electrons move slowly and thus interact strongly with one another. In turn, these interactions can lead to surprising phenomena, from new forms of superconductivity to unexpected optical effects with potential for fundamental technological advancements. This project looks beyond the flat-band approaches to explore other unique features of moiré materials, such as their multilayer structure and structural patterns. The research focuses on exploring three main directions: i) Plasmonics in moiré materials. Plasmons are formed by the synchronized motion of electrons that can carry energy at very high speeds, opening possibilities for ultrafast electronics and communication; ii) Interaction of these moiré materials with light, aiming to develop fundamental principles for better solar-cell design and new tools to probe hidden quantum properties of materials; iii) Developing a theory of superconductivity for these moiré systems by comparing superconducting behaviors in different multilayer systems. Alongside the research activity, this project includes an education component designed to broadly improve visibility, accessibility, and participation in the field of condensed matter. The education plan also includes a significant component dedicated to addressing stuttering in academia, aimed at increasing student participation and educational attainment levels. The main initiatives of the plan are: i) to develop a series of do-at-home experiments emphasizing condensed matter principles, ii) to develop a series of undergraduate-friendly events at the National High Magnetic Field Laboratory that will demonstrate the breadth of condensed matter research and help stimulate undergraduate student participation in research, and iii) to compile videos and resources that include examples of role models in academia who stutter and techniques for dealing with stuttering. TECHNICAL SUMMARY: A central feature of moiré systems are the flat electronic bands that promote electron-electron correlations. In addition to the flat bands, however, moiré materials possess other characteristics, such as the multilayer structure or the presence of a superlattice, with implications that go beyond flat-band formation. The project’s overarching intellectual goal is to outline a path to transforming the understanding of moiré materials beyond the conventional paradigm of flat bands promoting interaction effects. The research plan is organized into three thrusts: i) Plasmonics and electron response under high electric fields in moiré materials, which will focus on the exploration of moiré plasmonics, with emphasis on the role of superlattices and correlated effects; introduce a theoretical framework to describe plasmon bands; and propose a new scheme for launching 2D plasmons. ii) Nonlinear optical phenomena in moiré materials: This thrust will develop design principles to optimize photovoltaic response and provide theoretical foundations for demonstrating how optical probes can directly probe quantum textures of many-body states. iii) Multilayer moiré graphene as an analog of the “isotope effect”: One of the key outstanding questions in the moiré graphene field is the nature of the pairing mechanism. The PI will study alternating twisted multilayer graphene systems as a function of layer number to identify the pairing mechanism and formulate testable predictions for future experiments. Alongside the research activity, this project includes an education component designed to broadly improve visibility, accessibility, and participation in the field of condensed matter. The education plan also includes a significant component dedicated to addressing stuttering in academia, aimed at increasing student participation and educational attainment levels. The main initiatives of the plan are: i) to develop a series of do-at-home experiments emphasizing condensed matter principles, ii) to develop a series of undergraduate-friendly events at the National High Magnetic Field Laboratory that will demonstrate the breadth of condensed matter research and help stimulate undergraduate student participation in research, and iii) to compile videos and resources that include examples of role models in academia who stutter and techniques for dealing with stuttering. 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 · 2026-05

Project Summary/Abstract To curb the rising rates of suicide, interventions that directly target the causal mechanisms for suicidal behaviors (SB; i.e., suicide capability) are needed. Virtual reality (VR) suicidal decision scenarios are a valid and safe proxy for SB among at-risk individuals and have enabled causal examinations of suicide capability mechanisms. Intriguingly, these studies, including our preliminary data (N = 63), show that exposure to VR suicidal decision scenarios may evoke clinically meaningful reductions in suicide risk via suicide capability mechanisms. These findings converge with mounting evidence for the effectiveness of VR-based exposure in treating numerous psychiatric disorders. We consider a putative suicide capability mechanism that can be safely therapeutically targeted using VR suicidal decision exposure: motivational relevance of the suicidal decision process that facilitates the development of non-threat associations (i.e., inhibitory learning) with suicide ideation (SI) and the reintegration of suicidal deterrents (i.e., negative consequences of SB; reasons for living). This aligns with recent electroencephalography (EEG) findings that suicide attempters, but not ideators, show decreased sustained reactivity to threat/violence, reflecting an ability to volitionally dampen fear-inducing aspects of SI (i.e., suicidal deterrents), temporarily increasing their ability to approach SB. Our pilot EEG data (N = 28 suicidal adults) show VR suicidal decision scenarios engage the target mechanism of motivational relevance. This K23 proposal will evaluate VR suicidal decision exposure as a mechanistic intervention for SB and is designed to make the next critical steps in intervention development: (1) provide preclinical evidence for the feasibility, acceptability, and safety of the intervention in a transdiagnostic sample of 100 adults with recent SI and 50% with a suicide attempt history; (2) test whether the intervention, relative to treatment as usual, effectively engages the proposed mechanistic targets (neural; increases in late-LPP to suicide-related images and fronto-central gamma during the VR decision scenarios) and behavioral indicators (subjective accessibility of suicidal deterrents; behavioral orientation toward death/life) both in lab and in daily life; and (3) test whether change in the mechanistic targets and behavioral indicators account for intervention-associated change in clinical outcomes related to SB. We will use a multi-method approach involving EEG, subjective and behavioral responses, ecological momentary assessment (EMA), and self-reports/interviews to assess changes pre-post the intervention. This K-award addresses crucial gaps in the candidate’s training and will prepare them to be a leader in suicide intervention science employing an experimental therapeutics approach. The candidate will receive advanced training from an expert mentorship team, including mentors Drs. Joiner, Siegle, and Scott and consultants Drs. Patrick and Krafty. Findings will inform future R-level studies focused on 1) intervention refinement and clinical translation and 2) integrative multi-method (neurophysiology, subjective, behavioral, EMA) approaches to further clarify suicide capability mechanisms and develop effective, mechanistically targeted, and scalable SB interventions.

NIH Research Projects · FY 2026 · 2026-05

PROJECT SUMMARY The HIV-1 capsid provides a protective conical enclosure for the transport of the viral genome to its integration sites inside the nucleus. The capsid must uncoat to release the newly formed vDNA and to establish a permanent infection of target cells. When? where? and how? capsids are uncoated remain unclear and hotly debated in the field. This is especially relevant, as new highly active antiretrovirals such as the clinically approved drug Lenacapavir are being developed to combat virus infection. For infection to proceed, the core must cross the nuclear pore complex (NPC) to deliver the viral DNA (vDNA) to integration sites in the nucleus. This process requires structural remodeling of the capsid, which adapts through elastic deformation to penetrate the NPC’s ~64 nm channel. However, the mechanisms underlying capsid remodeling remain poorly understood. My preliminary evidence suggests that the phase separation properties of the host factor CPSF6 stabilize capsid structures in vitro and facilitates capsid trafficking in the nucleus of living cells to nuclear speckles, which are actively transcribing chromatin compartments favored for HIV-1 integration. Disruption of these processes, including with capsid-targeting drugs, impairs nuclear entry and viral infectivity, highlighting their biological relevance. This proposal has two specific aims. AIM-1 will resolve how CPSF6 influences HIV-1 capsid morphology in vitro. Using affinity captured virus particles, correlative light and electron microscopy (CLEM), and cryo-electron tomography (cryo-ET), I will reconstruct CPSF6-bound capsid structure and test the morphological adaptations of the HIV-1 core by host-factor interactions. AIM-2 will extend these findings to infected cells. I will use live-cell imaging, and a CLEM-guided cryo-focused ion beam milling (FIB) of cells and cryo-ET of a lamella prepared at the location of HIV-1 cores to capture capsid structures at multiple stages of entry, from the cytoplasm, through NPCs, and into the nucleus near integration sites. This in situ approach will allow me to structurally map capsid adaptation and reveal how morphology evolves within distinct cellular environments. Together, these studies will define how host-cell interactions reshape HIV-1 capsids to promote its nuclear transport and facilitate integration. They will also provide insights into therapeutic development and identify vulnerable stages in the viral life cycle. This proposal also provides a rigorous interdisciplinary training environment. Under the mentorship of Drs’ Ashwanth Francis and Scott Stagg, and in collaboration with Dr. Owen Pornillos, I will build upon my training in live-cell HIV-1 imaging, and gain new additional expertise in CLEM, cryo-FIB, cryo-ET, and advanced structural image processing. I will build technical and computational skills needed for high-resolution analysis of structures pertaining to virus-host interactions. I will also engage in professional development through conference presentations, manuscript authorship, and mentoring undergraduate students. These experiences will prepare me for a career as an independent investigator at the intersection of structural virology, molecular biophysics, and cell biology.

NIH Research Projects · FY 2026 · 2026-04

Project Summary The induction of a host type I interferon (IFN-I) response is classically understood to play an essential role in combatting virus infections. Detection of microbial nucleic acids in the cytosol by pattern recognition receptors (PRRs) triggers a signal transduction cascade that culminates in the activation of the IFN-I response. Emerging studies have linked previously unknown roles for cytosolic DNA sensing PRRs in detecting aberrant DNA species of self-origin to trigger the onset of age-related diseases, neurological disorders, and autoinflammatory conditions. Thus, there is a critical need to delineate the mechanisms by which cytosolic DNA sensing PRRs control IFN-I activation to develop the next generation of therapeutics to treat viral infections and DNA damage driven inflammatory disease states. We have recently identified a novel cross-talk phenomenon between cytosolic DNA sensing PRRs and the non-canonical NF-κB signaling pathway. While the non-canonical NF-κB pathway primarily governs lymphoid organogenesis and B-cell survival and maintenance in response to extracellular ligation of select members of the TNF receptor superfamily, our data unexpectedly revealed that intracellular ligation of cytosolic DNA sensing PRRs also triggered non-canonical NF-κB signaling. Furthermore, non-canonical NF-ĸB pathway activation amplified the IFN-I response during cytosolic DNA sensing PRR stimulation. Further investigation revealed the central regulator of the non-canonical NF-ĸB pathway, NIK was critical in enhancing cytosolic DNA dependent activation of IFN-I, but surprisingly, did not require its signaling partner, IKKα, nor downstream non-canonical NF-ĸB signal transduction. Instead, we found NIK interacted with and activated STING, an essential signaling adaptor required for IFN-I activation downstream of cytosolic DNA sensing PRRs. The mechanisms by which NIK elicits STING signaling is incompletely defined while induction of NIK/non-canonical NF-ĸB signaling in the cytosolic DNA sensing pathway is unknown. Regulatory factors that control NIK signaling during cytosolic DNA sensing also remain poorly understood. The central hypothesis of this proposal is that the non-canonical NF-ĸB pathway activator, NIK, is induced via a previously unknown mechanism during cytosolic DNA sensing and upon stabilization, exerts an additional layer of control on IFN-I activation in the cytosolic DNA signaling pathway by supporting STING signal transduction events. Interrogating protein-protein interaction network datasets, we identified novel regulators of NIK-STING signaling and propose to investigate their contributions in our specific aims: (Aim 1) Determine how NIK/non-canonical NF-ĸB is activated upon cytosolic DNA sensing. (Aim 2) Decipher the mechanisms by which NIK engages STING signaling. We anticipate our studies will broadly impact both the STING biology and non-canonical NF-κB signal transduction fields and will shed new light into developing new therapeutic strategies in combating DNA driven inflammatory disease states.

NIH Research Projects · FY 2026 · 2026-04

Project Summary/Abstract In humans, rodents, and other mammals, the perinatal period of offspring development (i.e., gestational and lactational) is marked by significant neurodevelopment and plasticity. During this period, maternal nutrition strongly shapes the developmental assembly and later function of offspring neural circuits. Compelling clinical and pre-clinical evidence indicates that maternal consumption of "Western Diets" (WD) high in fat and sugar content promote offspring susceptibility to obesity, at least in part by increasing offspring preference for WD-type foods. Based on a strong foundation of published literature and pilot data obtained in our novel Gcg- Cre::tdTom reporter rat model, we propose that the deleterious effects of perinatal WD exposure in offspring occur, at least in part, as a consequence of developmental plasticity and persistently altered function of central glucagon-like 1 (GLP1) signaling pathways. We specifically hypothesize that perinatal maternal WD alters the structure and function of GLP1 axonal projections in offspring, thereby reducing endogenous GLP1R signaling in reward- related brain regions that control palatable food intake. To test predictions arising from this overarching hypothesis, we will (1) document the nature and persistence of perinatal WD- induced plasticity in central GLP1 signaling pathways in male and female offspring, (2) examine how alterations in the structure and function of central GLP1 signaling pathways are associated with behavioral and physiological outcomes, and (3) investigate whether adolescent-onset treatment with the brain-penetrant GLP1R analogue semaglutide (SEMA) abrogates the effects of perinatal WD on offspring behavior, physiology, and endogenous GLP1R signaling. By elucidating for the first time how perinatal WD exposure impacts the structure and function of the central GLP1 system, results from this project will provide new insights regarding how the early nutritional environment modifies food reward-driven behaviors and lifespan metabolic health. If our data support this hypothesis, the results will have implications for developing strategies to combat long-term health adversities associated with early life exposure to WD.

NIH Research Projects · FY 2026 · 2026-04

PROJECT SUMMARY/ABSTRACT Episodic memory allows us to project ourselves back in time to mentally re-experience the past in vivid detail. This remarkable feat helps to define our personal identities and enables context-appropriate predictions that guide adaptive behavior. Unfortunately, episodic memory loss is a common cognitive consequence of neurological disease, psychiatric disorders, and normative aging, that reflects dysfunction in the hippocampus and the neocortical network to which it connects. This decline can present along a range of severity that spans an impoverished ability to recollect fine-grained event details, e.g., the color of the shirt your spouse was wearing at a birthday party, to forgetting course-grained event details, e.g., having attended a birthday party last weekend. Treating these kinds of memory loss remains challenging because interventions must target information that is definitionally idiosyncratic. However, recent methodological and theoretical advances in cognitive neuroscience reveal novel strategies that may solve this problem. Evidence from behavioral, neuroimaging, and computational modeling studies suggest that the degree to which episodic memories overlap in the hippocampus can be modified in a predictable, experience-dependent manner. Specifically, strong concurrent reactivation of multiple memories can enhance recall of coarse-grained episodic information by increasing hippocampal integration. Conversely, coupling strong and moderate reactivation of memories can enhance recall of fine-grained episodic information be increasing hippocampal differentiation. Consistent with this idea, recent evidence suggests that using a smartphone to record and subsequently replay real-world memory cues can indeed improve episodic memory by reducing representational overlap among memories in the hippocampus. Against this background, this proposal is organized around two primary aims. First, we seek to establish smartphone-guided memory reactivation protocols that selectively promote integration and differentiation of real-world episodic memories in the hippocampus. Experiments in this aim will specifically ask whether unstructured and structured (in terms of narrative, space, and time) reactivation can be used to achieve these outcomes. Second, we will ask whether the benefits of memory reactivation are associated with behavioral and representational costs. Experiments in this aim will probe for undesirable downsides that accompany, or potentially explain, reactivation-based improvements in episodic memory. Our approach combines smartphone technology, pattern-based analysis of functional neuroimaging data, and careful characterization of recall data with neurocognitive theory inspired by computational models of learning mechanisms. Achieving our aims will establish an empirical foundation on which future interventions can be built to systematically target real-world episodic memories at a level of abstraction, i.e., fine-grained vs. coarse-grained detail, that is appropriate for the severity of memory loss.

NIH Research Projects · FY 2026 · 2026-03

Our long-term objective is to determine how thalamic networks contribute to sleep and memory in the normal and diseased brain. The goal of this project is to investigate the basic mechanisms by which the spike dynamics in cognitive thalamic circuits contribute to sleep stability and memory consolidation. A growing body of literature indicates that sleep fragmentation occurs in pre-clinical stages and predicts the progression of Alzheimer’s Disease (AD). Thalamocortical neurons have been implicated in the progression of AD and their sleep activity is also associated with increased sleep stability. Remarkably, the mechanistic link between thalamic spike activity, sleep stability, and AD progression remains unclear. We aim to answer two questions. First, we will use optogenetics, pharmacology, and electrophysiology in behaving rats to understand how the spike dynamics (such as spikes occurring in groups or bursts) in thalamocortical neurons regulate sleep microarchitecture and stability. Second, we will use a rat AD model to test the hypothesis that increasing the occurrence of spike bursts in thalamocortical neurons delays the progression of AD disease and the deterioration of memory. The outcome of this work will be a better understanding of the basic mechanisms and causal contribution of thalamocortical neurons to sleep stability. We will also determine how increasing sleep stability benefits memory function and delays AD progression in a rat model of AD. Given the common occurrence of sleep disruption in neurodegenerative diseases, these outcomes can significantly contribute to the development of effective therapeutic interventions for Alzheimer’s Disease and related dementias.

NIH Research Projects · FY 2026 · 2026-03

Healthy Choices is a four-session behavior change communication intervention that was developmentally tailored for emerging adults to address self-management of health behaviors and HIV with evidence of positive effect on stigma and depression, built on Motivational Enhancement Therapy, integrating Motivational Interviewing with brief cognitive-behavioral strategies. Healthy Choices can be delivered in community settings by trained community health workers. When delivered with fidelity and in adequate dose, Healthy Choices results in reductions in alcohol use, HIV stigma, viral loads, and depression over follow-up compared to standard care. In the Dominican Republic, stigmas harm young people with HIV (YPWH). The Dominican Republic is a low- to middle- income country in the Latin America and Caribbean region, is 1 of 5 countries that accounts for over 95% of Caribbean HIV infections and has a culture that perpetuates stigmatizing attitudes and behaviors toward YPWH. To our knowledge, there are no Spanish-language interventions that concurrently address mental health, viral suppression, and stigma, tailored for young adults who are in a developmental period marked by exploration and a need for autonomy in health decision making. Considering the need for stigma reduction among YPWH to improve rates of viral suppression, reduce poor mental health, and encourage healthy coping by reducing problem alcohol use, we propose to adapt Healthy Choices for Spanish with local contexts plus co-create implementation strategies with community advising for future scale up. We will pilot test the adapted intervention and proposed intervention strategies, using a community-led implementation approach for feasibility, acceptability, and to assess for a signal of potential effectiveness on continuum of care outcomes including antiretroviral adherence and viral load. Considering the importance of context and the community-orientation of this study, we will apply the Exploration, Preparation, Implementation, Sustainment (EPIS) implementation science model, focusing on the Exploration and Preparation phases. We will engage community-based organizations (CBO) and a clinic in the Dominican Republic that work extensively with YPWH. Our investigator team has a long history of successful collaboration with impact on public health policy in the Latin American and Caribbean region. We propose three aims: (1) Elucidate barriers and implementation strategies for the Healthy Choices intervention; (2) Adapt and culturally translate the intervention for local contexts, and (3) Pilot test Healthy Choices with implementation strategies for feasibility and acceptability. If successful, we will have preliminary data for a full-scale hybrid type 1 effectiveness implementation randomized controlled trial of the intervention for underserved Spanish-speaking YPWH in the Dominican Republic with potential relevance to Spanish-speaking groups in the United States.

NIH Research Projects · FY 2026 · 2026-02

Project Summary Maternal and infant health outcomes in the U.S. are notoriously poor compared to other high-income nations. Within the US, outcomes vary widely across states, with some states (e.g., Mississippi) having infant and maternal mortality rates more than twice the national average. Understanding the structural factors (i.e., legal, social, and economic contexts) that lead to worse maternal and infant health across US states is critical to reducing maternal and infant health inequities. Additionally, understanding the role of established perinatal risk factors (such as pregnancy intentions, prenatal care, intimate partner violence, and social and economic stressors) in pathways connecting contexts to health is imperative. Finally, because Black and other racial/ethnic groups experience much higher rates of maternal and infant morbidity and mortality relative to white women, it is vital to examine whether the associations between context and health are more pronounced among these groups. This study will directly address these gaps and innovate upon extant research by: 1) developing the most comprehensive and robust set of state-level context measures to date and making them publicly-available to accelerate the pace of scientific research, 2) conducting the first detailed analysis of the relationship between state-level structural factors and a wide range of maternal and infant health outcomes while adjusting for a variety of other individual covariates 3) investigating the mediating role of established perinatal risk factors, and 4) determining whether the relationship between structural factors and maternal and infant health varies by mothers’ race/ethnicity. To accomplish these objectives, we will leverage 20 years of data from the CDC’s Pregnancy Risk and Monitoring Study (PRAMS) (N= 765,809) births across 45 states over 20 years) to examine the relationship between structural factors and maternal infant health both across states and within states as they change over time. This study directly addresses the National Institute of Child Health and Human Development’s goal of identifying structural factors shaping maternal health and health care. The results of this study have the potential to substantially advance knowledge of understudied structural determinants of health and to generate critical insights for the development of future strategies to improve maternal and infant health.

NIH Research Projects · FY 2025 · 2026-02

PROJECT SUMMARY Arrhythmogenic cardiomyopathy (ACM) is an inherited heart disease and a leading cause of sudden cardiac death. ACM is plagued by arrhythmias, cardiac dysfunction, and myocardial inflammation; phenotypes recently shown to be driven, in part, by chronic immune signaling. Our lab utilizes a mouse model of ACM (Dsg2mut/mut), that recapitulates key phenotypes observed in patients with ACM. Importantly, cardiac dysfunction and arrhythmias precedes overt myocardial injury. Recently, our lab demonstrated ACM cardiomyocytes release numerous cytokines, chemokines and DAMPs (Chelko et al., JCI [2024] & STM [2021]), which act to recruit pro-inflammatory CCR2+ macrophages; resulting in extensive myocardial cell death, inflammation, and fibrosis. This remodeling leads to cardiac dysfunction, arrhythmias, and increased mortality. Almost certainly, immune signaling occurs prior to disease onset and overt cardiac remodeling in ACM. Neutrophils are among the first cells to participate in the immune response, such as the formation of neutrophil extracellular traps (NETs) and neutrophil degranulation. In this proposal, we aim to investigate the role of two neutrophil-specific inflammatory proteins: peptidyl arginine deiminase-4 (PAD4) and myeloperoxidase (MPO). Specifically, PAD4- dependent NET formation and neutrophil-mediated MPO release prior to disease onset and throughout disease progression in Dsg2mut/mut mice. Our prior published works and preliminary data presented here demonstrate MPO and PAD4 are both upregulated in adolescent Dsg2mut/mut mice before phenotypes are present. We hypothesize both PAD4 and MPO contribute to a cardiac inflammatory environment, triggering immune cell chemotaxis, myocyte cell death, fibrotic remodeling, and cardiac dysfunction. To directly test this hypothesis, (Aim 1) we will test the efficacy of the MPO inhibitor, PF1355, in Dsg2mut/mut mice; and (Aim 2) genetically ablate Pad4 in Dsg2mut/mut mice. Both approaches will elucidate whether neutrophil-mediated immune signaling drives ACM pathogenesis. Outcomes from our work will determine if targeting neutrophil-mediated pathways could mitigate disease progression, providing a new therapeutic strategy in combating immune-mediated myocardial injury in affected individuals. This project will be completed at Florida State University College of Medicine (FSUCOM) under the guidance of Drs. Stephen Chelko (Mentor) and Yi Ren (Co-mentor). The training plan has been formulated to facilitate the development of both technical and conceptual proficiencies to successfully execute the proposed aims. Additionally, the training plan incorporates essential elements to transition the applicant into an independent scientific investigator. The Chelko and Ren Laboratories, the Department of Biomedical Sciences at FSUCOM, and the FSU campus – writ large – provides a rich scientific environment and outstanding graduate training for the applicant.

NSF Awards · FY 2026 · 2026-01

Wildfires are becoming more frequent, intense, and difficult to manage due to the shift in the global climate and expanding development into fire-prone areas. A major challenge in wildfire response is predicting how fires will behave, especially in forested areas where wind, vegetation, and fire interact in complex ways. Current fire models oversimplify forests as static blocks and fail to capture how trees sway, bend, and influence airflow. These oversights can lead to inaccurate forecasts and limit the effectiveness of prescribed burns and emergency planning. This project seeks to change that by developing more realistic, science-based tools to help land managers better anticipate fire behavior. The research team will also engage with fire professionals, students, and educators to ensure that the science gained through this research is applicable in real-world settings. The broader impacts include improving public safety, enhancing wildfire resilience, and training a new generation of interdisciplinary wildfire scientists. This project will develop and validate a new predictive modeling framework that explicitly incorporates canopy biomechanics, aerodynamics, and fire-atmosphere interactions to capture the dynamic motion of real tree canopies and their effects on fire behavior. Using multi-physics modeling and controlled laboratory experiments, the team will simulate how flexible trees interact with wind and buoyant flows during surface fires. These models will incorporate realistic canopy geometries, fire-atmosphere interactions, and turbulence to better understand key processes such as the transition from surface to crown fires. A multi-fidelity modeling approach and high-quality experimental datasets will be used to reduce uncertainties and improve model accuracy. With this, the project offers a new framework to address current limitations in realistic prediction of understory fire behavior. The resulting tools will support proactive fire management, prescribed burn planning, and risk assessment in the wildland-urban interface. Beyond wildfire science, the project advances computational modeling, biomechanics, and environmental fluid dynamics, with potential applications in agriculture, climate resilience, and natural hazard preparedness. 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 · 2026-01

Scientific progress increasingly depends on sharing powerful machine learning models across stakeholders, especially in sensitive fields like medicine, genomics, and disaster response. Machine Learning as a Service (MLaaS) allows researchers to collaborate without directly exchanging proprietary machine learning models or private data. However, these systems face serious and growing security vulnerabilities. Adversaries can steal models, reconstruct sensitive inputs, or intercept private data in transit, putting years of investment and sensitive societal applications at risk. Current computing platforms lack the protections necessary to guard against such attacks, creating an urgent need for secure infrastructure that supports scientific collaborations without compromising trust. This project develops a security-focused framework to protect collaborative scientific computing in MLaaS environments, supported through a partnership with Florida’s regional data center serving educational and governmental organizations. The project also integrates education, mentoring, and outreach activities to grow the workforce capable of safeguarding future scientific innovation. The project consists of three main research thrusts. First, it develops robust model protection techniques that hinder reverse engineering of machine learning models while preserving their utility. Second, it introduces behavioral monitoring tools to detect and respond to misuse of models, safeguarding sensitive input data without disrupting legitimate scientific computing activities. Third, it enhances data privacy through encryption schemes that allow model usage without exposing user inputs or model inference results. All components are designed for seamless integration into existing workflows and infrastructures. Collectively, these thrusts target trust, safety, and accessibility in MLaaS-based scientific collaborations, and research findings can be widely disseminated through open-source tools, educational modules, and community partnerships. 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 2025 · 2025-12

The Labrador Sea, part of the North Atlantic Ocean between Canada and Greenland, is one of the few places in the world where the deep ocean is ventilated. It is where salts and dissolved gases enter the deep ocean, thereby setting the chemical and physical environments of the deep ocean. An earlier study using data collected during a field campaign in the Labrador Sea from fall 2016 to spring 2017 showed that the traditional mathematical formulas used to calculate air-sea gas fluxes, i.e., how fast gases go in and out of the ocean, are inaccurate for the Labrador Sea. The reason may have to do with the unique environment there: strong wind, big waves, and chilling air. The earlier study indicated that it is not possible to make accurate estimates of how much gases, such as oxygen and carbon dioxide, are supplied to the deep ocean. Since oxygen is crucial to marine animals and carbon dioxide is one of the greenhouse gases contributing to global warming, there is an urgent need to develop more accurate formulas for air-sea gas fluxes that could be used in the Labrador Sea and elsewhere. In this project, scientists will go to the Labrador Sea in fall 2023 to make detailed measurements of oceanic and atmospheric conditions near the air-sea interface using recently developed techniques such as autonomous vehicles. In addition, high-fidelity computer simulations, only possible using supercomputers, will be conducted for the physical and chemical environments of the upper ocean. By synthesizing the new data and computer solutions, more accurate mathematical formulas for air-sea gas fluxes suitable for the world’s oceans including the Labrador Sea will be developed. The overarching objectives of the proposed study are to better understand bubble processes and bubble-mediated gas transfer and to propose a revised parameterization suitable for the world’s ocean, including strongly convective environments typical of the high-latitude ocean. An associated objective is to quantify the effect of solubility on bubble-mediated gas transfer. The proposed research includes an observational program and a modeling program. The observational component is part of the Bubble Exchange in the Labrador Sea (BELS) experiment – an international program during the Fall of 2023 in the Labrador Sea. The researchers will measure bubble-mediated air-sea invasion rates of CO2/O2/N2 and evasion rates of 3He/SF6 and make detailed measurements of gas flux forcing including bubbles and turbulent currents in the mixed layer. They will employ various approaches including 1D- and 3D-budgets, shipboard direct eddy covariance fluxes, as well as autonomous vehicles. These observations will be synthesized using state-of-the-art numerical models that concurrently simulate turbulent ocean currents, bubbles, and dissolved gases. The project aims to improve parameterizations of air-sea gas fluxes and reduce uncertainty in future predictions of gas uptake during ocean ventilation that may result from global warming. 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 2025 · 2025-10

The current information era closely relates to the Internet technology with traffic projected to grow exponentially in years to come. Although there are many proposals on how to deal with the upcoming bandwidth capacity crunch, the security of optical networks seems to be almost completely neglected. By taping out the portion of a dense wavelength division multiplexing signal, huge amounts of data can be compromised. Therefore, the security of the future network infrastructure is becoming one of the major issues—to be addressed sooner, rather than later. In this project, the University of Arizona (UA) team will coherently utilize the concepts of cryptography, quantum information theory, and nanophotonics to develop the next generation of quantum-enabled secure communication networks. The proposed project will significantly contribute to the major effort of providing ultimate security for future information infrastructure in the US as well as globally. At the same time, the proposed high-speed, secure, reliable quantum networking approaches will be a framework for cross-disciplinary research in quantum networks, cryptography, quantum information theory, quantum nanophotonics, coding theory, and fiber-optics technologies. This project will advance the quantum information science and technology by formulating a new framework to enable high-rate, robust, and scalable terrestrial quantum communication networks (QCNs) that use novel hybrid continuous variable (CV)-discrete variable (DV) protocols to achieve multiaccess quantum key distribution (QKD). To extend the transmission distance between nodes, the project will pursue postquantum cryptography/covert channel-based error correction, restricted eavesdropping, and hybrid measurement-device-independent (MDI)-QKD concepts. The proposed QCNs will be highly robust against channel impairments, including dispersion effects in fiber links and atmospheric turbulence in free-space optical links. By simultaneously solving the existing problems in both DV- and CV-QKD schemes and advancing towards QCNs, the UA team will develop an innovative concept and framework to attain the ultimate security for future network infrastructure in the US. The project focus is to: 1) develop novel hybrid CV-DV QKD protocols with extremely high secret key rates (SKRs) on the order of 10s of Gb/s; 2) fabricate high-speed integrated transceivers to support the proposed hybrid CV-DV QKD schemes; 3) develop postquantum cryptography/covert channel-based error correction for the hybrid CV-DV QKD, the restricted-eavesdropping concept, and hybrid MDI-QKD to significantly extend achievable transmission distances and increase the SKR; and 4) design quantum networking architectures based on these novel QKD concepts and experimentally demonstrate the proposed QCN concepts in a new terrestrial prototype at UA. The proposed QCNs will be genuinely secured by the fundamental principles of quantum physics, with secret key rates comparable to the classical-communication network data rates. Moreover, the proposed QCNs will provide an unprecedented security level for technologies with major societal and social impacts and benefits, suchas 6G wireless networks, the Internet-of-Things (IoT), and autonomous vehicles. 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 2025 · 2025-10

NON-TECHNICAL SUMMARY This award supports fundamental research aimed at improving our understanding and predictive capabilities regarding the formation and degradation of high-entropy metal oxides (HEMOs) in aqueous environments. HEMOs typically consist of four or more metal elements in near-equimolar ratios and are at the forefront of innovation in energy technologies, including batteries, fuel cells, and catalysts. However, their synthesis and aqueous stability remains poorly understood. To address these challenges, the research team will utilize high-throughput computational modeling and machine learning to identify HEMO compositions that are both readily synthesizable and highly resistant to aqueous corrosion. The supported research will be tightly integrated with educational efforts through engagement of a broad audience (including high school and undergraduate students) through augmented virtual reality learning tools, online videos, and hands-on research experiences. The project will also provide open-access databases and open-source software tools to the broader materials science community. TECHNICAL SUMMARY The supported research will address a critical knowledge gap in understanding the aqueous-phase synthesis and corrosion behavior of high entropy metal oxides (HEMOs), a class of multicomponent materials with broad potential for energy storage and conversion applications. The immense compositional design space of HEMOs, combined with limited experimental and theoretical understanding of their precipitation and dissolution processes, presents a significant barrier to their scalable development. The research team will aim to: (i) Develop a universal machine learning model to accurately estimate the free energies of HEMOs at aqueous conditions; (ii) Quantify nucleation and dissolution rates of HEMOs in aqueous solution combining machine learning with nucleation theory; and (iii) Implement a kinetic Monte Carlo (KMC) simulation framework to model real-time precipitation and corrosion dynamics. These models established will also be calibrated and validated through collaborative experimental efforts as well as text mining of the experimental literature. The project will produce an open-source database, predictive tools, and simulation codes for high-entropy materials. Broader impacts will include efforts to engage students in diverse fields including data science, solid-state chemistry, energy storage and conversion, and nanoscience. To reach a broad range of participants, the project will integrate augmented and virtual reality applications into university curricula, engage K-12 students by collaborating with Young Scholars Program at FSU, and expand access to research content via a YouTube-based educational channel. These efforts aim to foster interest in materials chemistry and provide educational opportunities across a wide range of learning communities. STATEMENT OF MERIT REVIEW 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 2025 · 2025-10

The number of users affected by data breaches and cyberattacks in the United States has increased recently, making the country the most targeted in 2023. Concurrently, the evolution of the Internet of Things communications and machine learning algorithms has led to an unprecedented influx of data requiring analysis. This has intensified the urgency to identify efficient methods for safeguarding exchanged information during data communication and Machine-Learning training. Fully homomorphic encryption, operating on encrypted data without decryption, emerges as a promising solution. However, its practical integration faces challenges, notably in algorithm intricacy and computational constraints, especially regarding latency. This project aims to optimize resource-intensive operations in existing fully homomorphic encryption schemes and seamlessly integrate the optimized algorithm into federated learning frameworks. This will enhance security while preserving learning performance, revolutionizing secure data analysis in Internet of Things communications. By benefiting federated learning users and cloud providers, this project will enhance security in practical applications such as telehealth and wireless communications, contributing to enhanced privacy, security, and efficiency in critical sectors, ultimately advancing science for society. The project develops an optimized fully homomorphic encryption algorithm to enhance security in Internet of Things (IoT) communications and integrates the optimized fully homomorphic encryption algorithm into federated learning frameworks to enable a secure training process on the encrypted data while maintaining learning performance. The project encompasses the following objectives: (1) developing a low-complexity, fully homomorphic encryption algorithm by optimizing resource-intensive operations; (2) integrating the optimized fully homomorphic encryption algorithm into federated learning to bolster security in IoT communications; and (3) accelerating fully homomorphic encryption and federated learning using parallel processing on graphics processing units and harnessing the combined computational power of both central processing units and graphics processing units. The proposed methodology is a pivotal stride toward propelling secure IoT communications into the future. By synergizing the domains of homomorphic encryption, parallel processing, and machine learning, this approach advances the field’s theoretical underpinnings and introduces novel methodologies that contribute to its growth and evolution. 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 2025 · 2025-10

Understanding how to compare and interpolate complex data, such as images, shapes, and network structures, is a fundamental challenge across science and engineering, especially in the context of artificial intelligence. This project develops new mathematical and computational tools that extend the theory of optimal transport, a well-established framework for measuring distances between probability distributions. The proposed methods are tailored to settings which more closely reflect specialized real world data structures than those considered in classical optimal transport. These advances will enable more adequate quantitative analysis methods for medical images, dynamic crowd movements, and biological network structures. An integral outcome of the project will consist in the production of robust and open-source software packages, which will make these generalized optimal transport methods accessible to researchers and practitioners in biomedical imaging, machine learning, and network analysis. Importantly, these algorithms will be firmly grounded in mathematical theory. The project will also train graduate and postdoctoral researchers through cross disciplinary collaborations, foster community engagement via a workshop, and engage with the broader community via a coding-focused course and K-12 outreach activities. The project pursues three interlocking aims. First, it formulates a new Constrained Unbalanced Optimal Transport model for comparing general positive measures under integral and parametric constraints; this involves rigorous well posedness results and efficient numerical solvers, targeted at shape analysis and population/crowd modeling. Second, it introduces an Optimal Riemannian Metric Transport framework, which blends ideas from optimal transport and infinite-dimensional geometry to compare Riemannian metrics on a fixed manifold; this framework is anchored in geometric connections to the well-established Wasserstein-Fisher-Rao metric, which has been proven successful in applications in machine learning and data science. Third, it investigates the Alexandrov and Riemannian geometry of Gromov–Wasserstein distances; this will result in geometry driven computational tools for comparing structured data on different domains, with applications to the analysis of astrocyte cell morphology. Together, these efforts will yield new methodological insights and a suite of software libraries. 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 2025 · 2025-10

Project Summary. The prevalence of daily cannabis use and cannabis use disorder (CUD) has increased in the United States over the past two decades. Unfortunately, psychosocial treatments produce minimal long- term abstinence rates and no FDA-approved medications for CUD exist. Thus, identifying novel CUD treatment targets is an increasingly urgent public health need. Stress-elicited cannabis use motivation has been implicated in worse CUD outcomes, but a mechanistic understanding of how acute stress increases cannabis use motivation in CUD is limited. The PI’s work has demonstrated that acute psychosocial stress enhancement of subsequent cannabis cue incentive salience, as indexed by the late positive potential (neural measure of approach-motivated attention recorded using electroencephalography [EEG]), was associated with worse CUD severity and intervention response, independent of subjective craving. Moreover, hypothalamic pituitary adrenal [HPA]-axis, rather than noradrenergic or subjective reactivity to the psychosocial stressor was associated with subsequent potentiation of the cannabis cue-elicited late positive potential. These studies suggest that non-genomic, rapid glucocorticoid effects may be a contributing mechanism in stress amplification of neural drug-cue reactivity, but their correlational designs preclude causal inference. Further, psychosocial stressors are unable to isolate HPA-axis vs. noradrenergic components of stress reactivity. To isolate HPA- axis activation and test causality, pharmacological manipulations, common in animal models but rare in human studies, will be used to produce separate and co-operative glucocorticoid (hydrocortisone) and noradrenergic (yohimbine) activation. We will employ a 2x2 randomized, placebo-controlled double-blind crossover design in 36 cannabis users with severe CUD. Our primary aim is to test the causal potentiating effect of glucocorticoids on neural drug-cue reactivity, and further determine if the effect depends on co-occurring noradrenergic stimulation. Preclinical work indicates that glucocorticoids can potentiate reward motivation via mobilization of endocannabinoid activity (primary target of cannabis). Thus, as an exploratory aim in line with NIDA priorities (NOT-DA-22-048), we will obtain plasma samples to test the impact of pharmacological stress on circulating endocannabinoids and their mediating role in glucocorticoid potentiation of neural drug-cue reactivity. This project represents a highly novel integration of a rigorous pharmacological challenge design with biological markers of drug-cue incentive salience and endocannabinoid system activity. If hypotheses are confirmed, one causal mechanism through which stress increases neural cannabis cue reactivity will be known, which has immediate implications for testing experimental therapeutics. The long-term goal is to understand how a stress- related mechanism predictive of worse CUD phenotype is generated and can be blocked in CUD. Development of this model will provide a valid, efficient and (relative to other neuroimaging methods) low-cost approach to screen candidate medications and optimize psychosocial drug cue exposure therapies.

NSF Awards · FY 2025 · 2025-10

The project comprises a strategic collaboration between Florida A&M University (FAMU) and Florida State University (FSU) involving both research and education, aiming to develop capacity in genome research via quantum machine learning (QML) at both universities. Researchers at both universities plan to address current challenges by developing scalable, interpretable, Artificial Intelligence (AI)-based computational tools to interpret single-cell data for understanding the evolution of cancer via single-cell sequencing. Single-cell sequencing - such as single-cell DNA sequencing (scDNA-seq) and single-cell RNA sequencing (scRNA-seq) - has been used to explain and predict how cancer cells evolve. Yet, current phylogenetic tree-inference tools are not scalable to the thousands of cells sequenced by scDNA-seq. In addition, no existing methods can fully automate the process of identifying normal cells in scRNA-seq data, a key step before inferring the cancer evolutionary tree. Consequently, the three main objectives of this project are the following: (1) using quantum-inspired computing to increase the scalability of building a phylogenetic tree on scDNA-seq data; (2) building fully automated, interpretable, deep-learning tools to distinguish between tumor and normal cells on scRNA-seq data; and (3) increasing the number of students studying explainable AI and QML in genome data. Broader-impact aspects of the work include also dissemination of the project via publications as well as open-source code. The project lies at the intersection of AI, genome research, machine learning, and quantum computing. Applying quantum computing to scDNA-seq data is anticipated to advance the frontier of AI-driven analysis using quantum-inspired computers while addressing the critical scalability issue of inferring the phylogenetic tree. The development of interpretable deep-learning methods on scRNA-seq data is expected to reveal the key features that distinguish between tumor and normal cells. It is anticipated that the work will not only contribute to the development of fully automated, interpretable, and robust deep-learning methods for the genomics field but also apply to other domains wherein interpretable data analysis is required, leading to the development of AI-based tools that can be widely used by clinicians and researchers, ultimately contributing to more personalized and effective cancer therapies. In addition, the successful completion of the proposed research will facilitate enhanced collaborative research between FAMU and FSU, train the workforce on AI at FAMU, bring application-driven AI education to FAMU, and accelerate the pace of genomic discovery and cancer-related single-cell data. 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.