Johns Hopkins University

NIH Research Projects · FY 2025 · 2025-09

Amblyopia is the leading cause of poor vision in infants and children. It arises because of abnormal visual experience early in life, during a critical window of development. Leading causes for the disruption of normal visual development are strabismus (misalignment of the visual axes between the eyes), anisometropia (mismatch in refractive error between the eyes) and deprivation (physical obstruction of one or both eyes). Clinically, amblyopia is diagnosed as reduced visual acuity in one eye that is not due to physical reasons and cannot be corrected optically. Consequently, much attention has been paid to amblyopic changes in early visual stages like the primary visual cortex (V1), and to development of treatments that can correct the loss of spatial vision (most prominently patching the weak, amblyopic eye), However, there are also deficits in higher visual functions localized outside of V1. The extent of these deficits, especially at the neural level, remains poorly understood, and even less is known about whether amblyopia treatments can overcome deficits in higher visual functions. The overarching goal of this project is to establish the ferret as a new animal model for studying amblyopia from the perspective of higher visual processing. Ferrets have a complex visual system with established similarities to the primate. This includes our recent demonstration that ferret PMLS functions as a higher visual area specialized for complex motion functions like motion integration, similar to primate MT. At the same time, larger cohorts of ferrets can be tested than is possible in primates, important for a disorder that can be quite variable in its depth between subjects, and is even more variable in the outcome of treatments. To establish ferrets as an amblyopia model, the work proposed here focuses on solving two problems. First, Aim 1 establishes a robust protocol for inducing amblyopia in ferrets. To mimic the human condition as closely as possible, we plan to trigger amblyopia by generating anisometropia early in life through a lens mounted in front of one eye. Second, Aim 2 will focus on establishing chronic recordings from flexible polymer probes to support longitudinal tracking of neural changes during amblyopia treatments. Together, these Aims will generate an experimental platform for future improvements of amblyopia treatments in human patients.

NIH Research Projects · FY 2025 · 2025-09

PROJECT SUMMARY The Coronary Artery Calcium Consortium (CAC Consortium) is a large, multi-center, retrospective clinical cohort study funded from 2013-2015 by the NHLBI to study the cardiovascular and non-cardiovascular prognostic value of CAC scanning in clinical practice. Despite its productivity, a major limitation for the CAC Consortium is the relatively short follow-up, with vital status ascertainment last conducted in 2014. This limitation becomes more important in an era of 30-year risk prediction ushered in by the publication of the PREVENT equations, as there are no data on the implications of CAC scoring over the 30-year time horizon. We seek to leverage the agreement between the National Institutes of Health (NIH) and the National Death Index (NDI) to extend CAC Consortium follow-up, with important scientific implications for 30-year risk estimation in young to middle-aged adults (age 30-59). Under this proposal, a total of ~5000 patients aged 30- 59 will have 30-year follow-up, and ~30,000 patients will have at least 25-year follow-up, providing sufficient statistical power for our scientific aims. In Aim 1, we see to quantify 30-year survival in patients aged 30-59 with baseline CAC=0. We hypothesize the older patients will gain a great survival advantage compared to younger patients compared to the general age-matched population. In Aim 2, we seek to describe the 10- and 30-year prognosis of individuals aged 30-59 with low absolute CAC scores between 1 and 100. While these patients have a low absolute CAC score, their percentile score is high, which may correlate with similarly poor survival to old age compared to older patients with higher absolute CAC scores. In Aim 3, we seek to evaluate the improvement in 30-year cardiovascular mortality risk prediction when CAC is added to the PREVENT equations. This extended follow-up of the CAC Consortium will leverage previous NHLBI investment in this project and fill critical knowledge gaps that could not be addressed with any other dataset, with specific implications for the critical age range (30-59 years old) where guidelines recommend risk assessment over a 30-year time horizon. We believe this work will have specific implications for future ACC/AHA Prevention Guidelines, which will likely incorporate recommendations based on long-term risk prediction.

NIH Research Projects · FY 2025 · 2025-09

Xylazine is a veterinary sedative that has no approved human use but that has, over the past half-decade, become increasingly prevalent in the United States’ illicit drug supply, often mixed with fentanyl. In a recent warning to stakeholders, the FDA identified risks associated with xylazine use, including that: 1) xylazine can cause respiratory depression and overdose which cannot be reversed by naloxone; 2) xylazine cessation can lead to severe withdrawal, potentially impeding treatment for opioid use disorder, and 3) xylazine use has been linked to wounds and severe necrotic skin ulcerations. There is an urgent need for rigorous epidemiologic data on xylazine use and associated health effects to inform public health response, treatment, and harm reduction. However, the epidemiology of xylazine use and its associated adverse health outcomes such as wounds and overdoses are poorly characterized. Our pilot study will provide among the first rigorous estimates of the association of xylazine use with adverse health outcomes, and will examine drug use behaviors that may exacerbate or mitigate these xylazine-related harms. Our aims specifically focus on three hypothesized xylazine-associated adverse health outcomes: 1) wounds and skin necrosis; 2) drug overdose; and 3) all- cause emergency and hospital visits and the hospital bills incurred because of those visits. These aims will be accomplished with infrastructure of the ALIVE study: a 30+ year observational cohort of community-recruited adults living in the Baltimore area who have a history of injection drug use (7,8). Enrolled participants attend twice-annual study visits at which they complete questionnaires assessing substance use and health. Beginning in 2025, urine samples will be collected from all participants for rapid, on-site xylazine and fentanyl testing. Linkage to Maryland’s Health Information Exchange (HIE) allows complete ascertainment of hospital- and emergency-department based healthcare utilization including all associated dates, locations, and diagnosis and procedure codes. To this existing infrastructure, we will add: 1) a supplemental questionnaire with detailed questions about xylazine us; and 2) a standardized direct-observation wound assessment protocol developed in collaboration with an expert on wound care for people who use drugs. If successful, our pilot will facilitate future research examining trends in the epidemiology of xylazine, including the prevalence and correlates of use and health consequences, as well as harm prevention and treatment.

NSF Awards · FY 2025 · 2025-09

Cells use one third of their energy to produce proteins. Thus, cells require to precisely control protein production. Current knowledge of protein production regulation is based on mixing the contents of millions of cells to measure the average protein production rates. However, we now know that individual cells can behave differently than population averages. This research program will develop and apply new methods that allow measuring protein production in individual cells to determine when and where the regulatory events happen. These methods are based on measuring the speed of individual ribosomes (the protein production machines of the cell) in live cells. Evaluating thousands of ribosomes allows discerning when and how protein production, as well as other key functions of ribosomes such as UV or high temperature responses. More broadly, the research team will introduce high school students to the use of microscopy to study cellular and molecular biology processes. It will also develop courses to provide the basic skills needed for imaging of live cells, and how to use these skills to further our understanding of the fundamental principles of cellular and molecular biology. Live-cell biosensors have revolutionized our understanding of signaling and cell cycle regulation by enabling biochemical characterization of individual cells as they respond to environmental perturbation. However, many important biochemical parameters remain inaccessible through live single-cell approaches. Single-molecule tracking (SMT) approaches have allowed researchers in the chromatin regulation field to quantify the amount of DNA-bound vs. free proteins in live single cells. This is because proteins diffuse slowly when they are bound to DNA. The PI’s team has recently discovered that many other proteins, including translation factors and signaling proteins, show qualitatively distinct diffusion properties depending on the size and properties of the complex they are bound to. Thus, measuring the diffusion properties of thousands of single-molecule trajectories in live cells allows to qualitatively increase the number of biochemical parameters that can be measured at the single cell level. The specific objectives are: 1) Dissect global translation dynamics in live single cells, 2) Dissect global translation dynamics during UV radiation-induced ribotoxic stress response (RSR), and 3) Implement an educational plan to (a) expose high school students to imaging based cellular and molecular biology ,and (b) develop a graduate course elective to provide the basic skills and knowledge required for live cell imaging based molecular and cellular biology. 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 · 2025-09

PROJECT SUMMARY/ABSTRACT Osteoporotic fractures increase morbidity and mortality in people with CF; vertebral fractures are the most common osteoporotic fracture in people with CF. Diagnosing CF osteoporosis is hindered by two critical gaps: the current standard of care (use of dual energy X-ray absorptiometry, DXA) has poor sensitivity for fracture in people with CF and fracture risk strata have not been defined in this population. Beyond DXA, other genetic and imaging data are used to predict fracture risk in people without CF. Genetics explains 50% of variation in vertebral fractures; genetic modifiers alter non-CF osteoporosis risk, but genetic modifiers specific to CF osteoporosis have not yet been identified. Volumetric bone density (vBMD) by CT is superior to bone density by DXA in predicting vertebral fracture in non-CF populations, but has not yet been investigated in people with CF. Identification of genetic modifiers associated with CF osteoporosis and building prediction models for osteoporosis using quantitative measures of bone strength derived from CT imaging can address this critical knowledge gap in CF osteoporosis. To test the hypothesis that genetic modifiers and imaging biomarkers will identify people with CF at highest risk for vertebral fracture, I will conduct three aims: 1) discover genetic modifiers of CF osteoporosis through candidate and genome wide association study approaches, 2) examine the relationship of volumetric bone density and vertebral fracture in people with CF, and 3) explore the association of genetic modifiers, volumetric bone density and vertebral fracture. Data generated will enable identification of people with CF at highest risk for vertebral fracture and be preliminary data for future R01 applications. This proposal will support the training of Dr. Malinda Wu, a pediatric endocrinologist, who is dedicated to an academic career studying osteoporosis, using CF as the model. During the K23 award period, she will acquire requisite skills in clinical and translational investigation through formal coursework and workshops, attendance of conferences and seminars, personalized mentoring and hands-on research experience. Dr. Wu will be mentored by Dr. Scott Blackman, expert in genetic modifiers of CF complications, and Dr. Janet Crane, expert in bone biology. She has assembled a diverse team of experts with the skills, resources, and expertise to guide her to successfully complete this project. In summary, the research and training plan proposed in this K23 application will broaden our understanding of CF osteoporosis which has implications for non-CF osteoporosis and provide the necessary training and investigational niche for Dr. Wu to develop a successful, independent career in research.

NIH Research Projects · FY 2025 · 2025-09

Abstract Invasive fungal pathogens cause 1.5 million deaths a year and Cryptococcus neoformans is the primary cause of fungal meningitis worldwide. Cryptococcus is a ubiquitous fungus that is inhaled from the environment and, through mechanisms that we still don’t understand, disseminates out of the lung and subsequently enters the brain. The Cryptococcus yeast morphotype has been widely studied, and the study of its anti-phagocytic capsule has been critical to combating this pathogen; however, the yeast cell type is not the only morphotype to consider in cryptococcal disease. Cryptococcus produces dormant and stress resistant basidiospores (sexual spores) that are smaller and better aerosolized than yeast and thus more likely to reach the lower airways. Importantly these spores have a distinct surface to yeast, lacking the anti-phagocytic capsule. Spores are the presumed infectious morphotype, yet due to difficulties associated in working with spores, the Cryptococcus spore surface remains undefined and relatively few studies exist evaluating spore-host interactions. Critically, Cryptococcus spores can can disseminate out of the host lung better than yeast which translates to spores of otherwise avirulent yeast causing disease in intranasal murine models of cryptococcosis. This preferential dissemination and disease are likely a result of spores being able to invade host lung cells better than yeast. As spores germinate into yeast, and become more yeast-like, this preferential internalization diminishes probably due to surface epitopes being masked as the yeast capsule is formed. The spore surface components that drive host cell invasion, dissemination and disease remain a mystery. This research, conducted in the Casadevall lab, will be centered on defining the distinct surface components of spores, determining which components drive preferential host cells invasion, and identifying the molecular mechanisms enabling Cryptococcus dissemination and disease. Previous work has shown that cell surface hydrophobicity plays a role in yeast phagocytosis. Additionally, spores have distinct surface sugar epitopes to yeast, suggesting a unique surface glycoproteome. In specific aim 1, the cell surface hydrophobicity of spores will be systematically perturbed and evaluated, and the spore surface proteome will be defined. These surface components will subsequently be probed for roles in host cell invasion. Once surface properties that drive host-cell invasion have been identified, the molecular mechanisms driving dissemination and disease can be identified. In specific aim 2, the role of spore internalization by specific host cell types in disease kinetics will be determined by identifying fungal surface components and host receptors that drive uptake by different resident lung cells. Next, the role of these interactions in murine models will be probed for changes in in vivo host cell invasion, dissemination and disease. This work will identify novel mechanisms of host cell invasion that drive dissemination and disease, which can subsequently be used in the development of novel diagnostic tools and treatments for invasive fungal diseases.

NIH Research Projects · FY 2025 · 2025-09

Project Abstract This project aims to assess the potential of pregnenolone, a signaling specific negative allosteric modulator (NAM) of the CB1 cannabinoid receptor, to reverse symptoms of cannabis intoxication. As defined by the DSM- 5, cannabis intoxication involves neuropsychiatric symptoms such as anxiety and impaired motor coordination and physiological symptoms including dry mouth and tachycardia arising shortly after cannabis use. These symptoms can be distressing and associated with significant clinical complications such as car accidents, cardiac arrhythmias, and psychotic episodes. Cannabis’ increasing popularity and potency has seen a surge in emergency room (ER) visits for management of intoxication, yet there are no medications indicated for the treatment of cannabis intoxication. NAMs of the CB1 receptor—the cannabinoid receptor whose activation by tetrahydrocannabinol (THC) facilitates intoxication—have shown promise as novel treatments for cannabis intoxication. Pregnenolone, an endogenous neurosteroid often marketed as a supplement, has been shown to act as a CB1 receptor NAM specifically attenuating the effects of THC. Preclinical studies demonstrating pregnenolone’s capacity to treat cannabis intoxication symptoms have led to interest in applying its mechanism to drug development yet its ability to reverse active intoxication symptoms have not yet been studied in humans. We will therefore set up a within-subjects, double-blind, placebo-controlled human laboratory study investigating whether pregnenolone can treat cannabis intoxication symptoms. Healthy, cannabis-naïve individuals who meet inclusion criteria will be randomized to one of four conditions: cannabis brownie and placebo capsule, cannabis brownie and 250 mg pregnenolone capsule, cannabis brownie and 500 mg pregnenolone capsule, and placebo brownie and placebo capsule (all cannabis brownies will contain 25 mg THC). Following study drug administration, participants will complete questionnaires evaluating subjective effects (via the Drug Effect Questionnaire), psychotomimetic symptoms (via the Positive and Negative Syndrome Scale) and cognitive/psychomotor impairment (via a battery of cognitive assessments). Heart rate will also be assessed. Blood will also be collected at various timepoints for later measurements of serum pregnenolone, THC, and THC’s metabolites 11-OH-THC and THCCOOH to assess any dose-dependent effects of pregnenolone on cannabis intoxication. We hypothesize that pregnenolone administration will be associated with significantly fewer cannabis intoxication symptoms compared to placebo as evidenced by lower scores on the questionnaires, and that higher levels of pregnenolone will also be associated with fewer subjective effects and signs of impairment associated with cannabis intoxication. Results from this study will provide information on whether agents that act as CB1-NAMs can serve as rescue therapies for cannabis intoxication symptoms, while also giving insights into whether pregnenolone itself could fulfill this role.

NSF Awards · FY 2025 · 2025-09

Computer proof assistants, software programs that verify the logical reasoning of mathematical proofs written in precise formal language, offer an exciting new paradigm for mathematical research, enabling large-scale collaborations and empowering individual researchers to interact with theorems in other subfields that can be found in large open-source libraries of formalized mathematics. Computer proof assistants are likely to become more integral to the working life of mathematicians in the future. As different proof assistants often implement the rules of different formal systems, mathematicians may even elect to prove certifiably-correct theorems in a non-standard “synthetic” foundation system. The PI will develop three project clusters that involve computer formalization of higher category theory in parallel with theoretical projects necessary to facilitate this work. The formalization projects each have several student collaborators, who are developing their skills with these tools while making significant scientific contributions. The first project will introduce (∞, 1)-category theory to Lean’s otherwise broad-ranging library Mathlib via the formalism of an ∞-cosmos, developed in prior joint work of the PI. This approach will leverage Mathlib’s existing bicategories library and largely sidestep the apparent difficulty in formalizing proofs that directly deploy the quasi-categories model. A second project also aims to formalize theorems about (∞, 1)-categories, but in a non-standard foundation system designed to make conceptually-simpler constructions and proofs fully rigorous. There is an experimental computer proof assistant Rzk that verifies proofs written in simplicial homotopy type theory, a formal system developed in prior joint work of the PI, and ongoing work to formalize (∞, 1)-category theory in this synthetic framework. This project aims to expand this system, which is insufficiently expressive to encompass the full theory of (∞, 1)-categories, without narrowing its semantics, which include (∞, 1)-categories defined internally to an arbitrary ∞-topos. The final project aspires to develop a prototypical synthetic theory of (∞, n)-categories by developing a new type theory for marked shapes to provide a finitary syntactic encoding of (∞, n)-categorical data with semantics in the new comical spaces model. These pen-and-paper developments will enable (∞, n)-category theory to be formalizable in the future, once a suitable computer proof assistant is built to implement the rules of this new proposed formal system. The medium-term objectives include specific formalization targets that will then be available to other users of Lean’s Mathlib and Rzk’s simplicial homotopy type theory libraries. The long-term objective is to make (∞, 1)-category theory and eventually also (∞, n)-category theory more accessible to non-experts, who could use the computer formalized proofs as ingredients in their work. 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 · 2025-09

MicroRNAs (miRNAs) are a conserved class of small, endogenous noncoding RNAs that regulate gene expression post-transcriptionally. They play critical roles in essential biological processes, including animal growth and development. Dysregulation of the miRNA pathway has been implicated in the pathogenesis of various human diseases. miRNAs exert their function through the miRNA-induced silencing complex (miRISC), which comprises the Argonaute protein (ALG-1 in C. elegans) and other protein cofactors. The composition of miRISC—and thus its ability to degrade mRNA targets—varies based on tissue type and physiological conditions. For example, somatic and germline miRISC recruit distinct cofactors, leading to different outcomes for target mRNAs. miRISC composition also shifts dynamically in response to environmental stress or aging. Under adverse environmental conditions, C. elegans enters an alternative, stress-resistant developmental stage known as dauer. A key unanswered question in the field is how the miRNA pathway rapidly transitions from regulating pro-developmental gene expression to stress-adaptive gene expression during this stage. Our preliminary studies reveal two novel protein interactors of miRISC during dauer: the conserved RNA-binding protein UNK-1 (Unkempt) and its binding partner, CRI-1 (Headcase). I hypothesize that UNK-1 and CRI-1 are critical for stress adaptation in dauer by interacting with miRISC to promote the degradation of pro-developmental gene targets. I will test this hypothesis in three specific aims: In Aim 1, I will characterize the interactions between UNK-1, CRI-1, and miRISC, and determine whether they are mediated by direct protein-protein interactions or indirectly through a shared set of mRNA targets. In Aim 2, I will identify the mRNA targets of UNK-1 and assess their overlap with miRISC targets. Preliminary data indicate that UNK-1 expression is highly upregulated during the dauer stage. In Aim 3, I will investigate the regulation of UNK-1 and CRI-1 expression during dauer and evaluate their roles in stress adaptation. The conservation of UNK-1 and CRI-1 across species, including humans, suggests that their role in miRISC regulation during stress may be evolutionarily preserved. This work could provide insights into fundamental processes such as metabolism, aging, and cancer, offering broader implications for understanding stress adaptation in complex organisms.

NIH Research Projects · FY 2025 · 2025-09

PROJECT SUMMARY Cranial neural crest cells are a population of multipotent stem cells essential for craniofacial development that contribute to the sensory ganglia and craniofacial skeleton. Neural crest cells are specified during neurulation before undergoing an epithelial-to-mesenchymal transition (EMT) and migration; while we appreciate the complex network of signaling pathways and transcription factors underlying specification and migration, the regulation of these events remains incompletely understood. We have recently identified that expression of the gene encoding Raftlin-2 is upregulated in neural crest cells during specification and maintained throughout migration. Raftlin proteins localize to the plasma membrane where they are reported to organize lipid raft domains and regulate diverse cell signaling events, leading us to hypothesize that Raftlin-2 contributes to the development of neural crest cells through an impact on cell signaling. In preliminary studies probing Raftlin-2 function in the cranial neural crest of avian embryos using multiple knockdown strategies, we observed reduced expression of several genes required for neural crest specification, suggesting that Raftlin-2 is essential for the establishment of the neural crest domain. Aim 1 will investigate the role of Raftlin-2 in regulating specification by using immunohistochemistry to measure neural crest specifier genes expression. Since specifier gene expression is initiated by cell signaling pathways including Wnts and Bone Morphogenetic Proteins, I will next use signaling activity-sensitive reporter constructs, and fluorescent in situ hybridization to investigate if Raftlin-2 regulates the activity of these pathways to govern specification. Finally, I will use RNA sequencing in Raftlin-2 deficient cranial neural crest cells to determine additional targets and signaling pathways dependent on Raftlin- 2 function. Our preliminary results also showed reduced neural crest migration area after Raftlin-2 knockdown compared to controls in vivo, and we find that Raftlin-2 loss of function resulted in decreased migratory persistence ex vivo, which led us to hypothesize that Raftlin-2 is necessary for effective neural crest migration and adhesion after specification. Aim 2 will test this hypothesis using ex vivo explants to assay neural crest cell directional migration and focal adhesion dynamics during chemotaxis. I will also use quantitative western blot analysis to probe the activation states of molecular effector proteins downstream of chemotactic signals to define and subsequently test the precise mechanisms of Raftlin-2 function during migration. Together these experiments will uncover how Raftlin-2 impacts developmental signaling pathways necessary for cranial neural crest cell specification and migration. These results will advance our understanding of how membrane organization regulates craniofacial development insight and will provide novel insights that can be used to treat craniofacial anomalies.

NIH Research Projects · FY 2025 · 2025-09

Project Summary Cells must adapt to shifting environmental conditions to ensure survival. And, while we know that many of these insults target the ribosome and protein synthesis, we have recently begun to appreciate the extent to which such cellular insults are directly sensed by ribosomes themselves to initiate stress signaling pathways. Pioneering studies established that the status of translating ribosomes is monitored to identify signs of global translational distress leading to activation of the several signaling pathways including the ribotoxic stress response (RSR) and the integrated stress response (ISR). Normally, the orchestration of translation initiation, elongation, and termination rates enables maintenance of ribosome equilibrium on messenger RNAs (mRNAs). However, cellular insults leading to the accumulation of damaged mRNAs inevitably result in prolonged stalling of ribosomes within coding sequences and eventually to ribosomal collisions which have been shown to be a key determinant for signaling translational distress. Low-level collisions prompt a ribosome-mediated quality control, while abundant transcriptome-wide collisions overwhelm the ribosomal rescue mechanisms and serve as a platform for initiating general stress response pathways. Our previous studies showed that ribosome collisions recruit and activate at least two cellular kinases, GCN2 and ZAK, triggering two distinct downstream signaling cascades. In this proposal, we propose to define molecular mechanisms for how these two kinases, GCN2 and ZAK, interact with ribosomes to activate downstream signaling. We will use primarily biochemical approaches both in vivo and in vitro to identify binding partners of these proteins and specific interactions critical to their output along the lifetime of their activity. Our in vivo focused studies will primarily rely on mass spectrometry approaches at low and high resolution. Our in vitro focused studies will evolve around a reconstitution approach using purified proteins and diverse ribosome populations. The centrality of these pathways in determining cell fate makes them critical to an understanding of cellular homeostasis in health and disease.

NSF Awards · FY 2025 · 2025-09

An entirely new perspective on the Universe emerged with humanity’s first observations of gravitational waves a decade ago. These messengers of gravity are ripples in the fabric of spacetime launched from the extreme environments of colliding black holes and detected on Earth for the first time in 2015 by the Laser Interferometer Gravitational Wave Observatory (LIGO). Over the past decade astronomers have used this gravitational messenger, along with traditional astronomical observations of light, to uncover many long-held secrets in physics and astronomy. We have, however, only just begun to learn about our Universe from this new discipline of multi-messenger astronomy. In only the last few years we have begun to break open a new realm of gravitational wave observations from merging black holes that are up to billions of times larger than the merging black holes detected by LIGO. These supermassive black holes probe different physical regimes and stand to again open the flood gates of discovery. This research program will build-up necessary theoretical frameworks needed to describe the astrophysical processes which generate multi-messenger signatures from these monstrous black hole mergers. The aim is to predict observational signatures of these supermassive black hole binaries interacting with their environments. To do so the program will pioneer new models of these interactions to facilitate their discovery and interpretation. In this way the program will advance the state-of-the art in this quickly moving field of multi-messenger astrophysics. It will furthermore contribute to education and training of a graduate student and undergraduate students. Supermassive black holes are the focus of this research program and the subject of a long-standing mystery: will they merge? What can that tell us about the environments that shape them? What are the consequences for the low-frequency gravitational wave sky that is now being probed by the Pulsar Timings Arrays, and in the future by the LISA observatory? The research program will advance the state of the art in modeling binary+gas interactions by laying the theoretical groundwork for a so-far-unexplored regime of binary accretion: hot accretion flows, as opposed to standard thin disk accretion. This program will utilize results established for single-black-hole hot accretion flows, apply them to the binary problem, and make much needed predictions for multi-messenger signatures of accreting supermassive binaries. 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 · 2025-09

Schizophrenia (SCZ) is a chronic neuropsychiatric disorder that affects 20 million people worldwide and is among the most burdensome of all health problems, with enormous implications for individual health, quality of life, and societal costs. Therefore, elucidating disease mechanism with the ultimate goal of developing improved treatment strategies is an urgent priority. Though SCZ is highly heritable, no singular genetic cause has been found and its genetic architecture is complex. A 2022 large genetic study identified rare genetic variants that confer substantial risk for SCZ. Among the 10 most significant is GRIA3, a gene on the X chromosome which encodes GluA3, the 3rd subunit of a major receptor mediating excitatory signals in the brain, the amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR). AMPARs bind the neurotransmitter glutamate and are key components of the cellular mechanisms enabling the encoding, storage, and integration of information throughout the lifetime of an organism. The long-term goal of this proposal is for the principal investigator (PI) to develop as an early independent investigator at the intersection of Neuroscience and Psychiatry, and to leverage training with experts in basic neuroscience to better understand the pathological mechanisms underlying psychiatric disorders such as SCZ with the ultimate aim to identify new therapeutic targets. The major objectives of this proposal are to undergo training in concepts and techniques in basic neuroscience to investigate the functional consequences of SCZ-associated GRIA3 variants in vitro and gene expression changes in Gria3-knockout mice in vivo. The central hypothesis is that SCZ-associated variants in GRIA3 result in partial or complete loss of synaptic GluA3, and alter AMPAR assembly, trafficking, and kinetics, resulting in multiple downstream effects that are phenotypically relevant to SCZ. Aim 1 is functional characterization of SCZ-associated GRIA3 variants, including missense changes as well as those predicted to abrogate protein expression, in order to evaluate effects on RNA and protein levels, AMPAR assembly, GluA3-cell surface delivery, spine-trafficking, kinetics, and binding to interacting proteins, and synapse number and structure, and morphology in neurons and heterologous cells in vitro. Aim 2 is performing single cell transcriptomics of forebrain and various brain regions in Gria3-null mice of both sexes at 1 and 3 months of age followed by validation studies. Expected outcomes of the proposed work are elucidation of the consequences of SCZ-associated GRIA3 variants on AMPAR receptor structure and function in vitro and single-cell transcriptional changes associated with GluA3 loss in mice. In addition to the described Aims, this proposal implements a structured, individualized career development plan specifically designed to support the candidate’s career objectives and facilitate a transition to independence. This plan will involve the guidance of an expert mentoring team, course work, supervised grant writing, and hands-on technical training, all within a collaborative, rigorous, successful academic environment.

NIH Research Projects · FY 2025 · 2025-09

More than 50% of people living with HIV (PLWH) encounter cognitive dysfunction, and chronic peripheral pain, in the setting of opioid drug abuse. Neuronal circuitry innervating from the prefrontal cortex to the striatum is important in decision-making. Additionally, dopaminergic projections from the ventral tegmental area (VTA) in the midbrain to the nucleus accumbens (NAc) and dorsal striatum are involved in reward and motivational behaviors. We reported that global suppression of secreted phosphoprotein-1 production (OPN/Spp1) increases the expression of the mitochondrial translocator protein (TSPO) in Iba1+ macrophages/microglia across several key brain regions involved in cognition including the substantia nigra (SN). Interestingly, a subset of tyrosine hydroxylase (TH) reactive neurons co-labeled with TSPO, or were closely positioned near TSPO+ TH- cells in the midbrain region. In this regard, substance use disorders (SUD) in PLWH and in particular methamphetamine (METH) continues to be a consequential comorbid condition impacting viral suppression and health outcomes. Findings using rat and mouse HIV-Tat transgenic models have provided insights about how METH alters the expression of genes required for dopamine synthesis, metabolism, and receptor trafficking, and the role of sex as a modifier. However, our understanding of how dopaminergic neuronal-glial communication is altered in vivo during HIV replication, and METH is not understood likely due to the daunting task of deconvoluting multiple intersecting variables. Moreover, we implicated mammalian target of rapamycin (mTOR) pathway activation by (OPN/Spp1) in a mechanism of neuroprotection. Whether mTOR-OPN/Spp1 signaling plays a role in microglial- dopaminergic neuronal crosstalk in HIV-METH infection-exposure in vivo in is unknown. In this R21 application for high-risk/reward ideas, we propose to use our expertise with HIV-infected humanized mice, SUDs and modeling, to develop a rational approach to mathematical model dynamic dopaminergic neural-glial network circuitry. Our overarching hypothesis is that neuro-glial cells upregulate OPN/Spp1 expression in response to HIV-1 infection, which stimulates mTOR pathway signaling thereby, activating neuroprotective signaling to preserve homeostatic neurocircuitry; with acute co-exposure to METH, these pathways are upregulated and reinforced in a time-dependent manner. We will interrogate gene expression among neurons and glia in the nigrostriatal and meso-limbic brain regions to resolve dopaminergic neuronal-glial interactions using prospectively collected in vivo time course data. The analyses will focus on identifying interactions in the presence of HIV infection, with and without ART and acute administration of METH. Our long-term goal is to gain insights into the therapeutic potentials of resilience from cognitive dysfunction and drug addiction. We expect that findings from this project will advance the understanding of immunomodulation and metabolic reprogramming during co-exposure to HIV-1 and METH and will provide avenues for translational and clinical research aiming at improving mental health and substance use.

NIH Research Projects · FY 2025 · 2025-08

PROJECT SUMMARY/ABSTRACT Candidate’s Long-Term Career Goals: By pursuing the training plan outlined in this F32 proposal, Dr. Andrew Davis will learn key skills in clinical epidemiology, causal inference, and informatics while writing a K23 award and ultimately becoming an independent health services researcher investigating evidenced-based interventions to improve outcomes for survivors of pneumonia and respiratory failure. Clinical Problem to be Addressed: Community acquired pneumonia (CAP) with hypoxemia remains a major driver of hospitalizations, readmissions and new post-discharge morbidity. Current CAP guidelines are limited and make no recommendations on best practices to support the safe discharge of patient’s home. Candidate Background: Dr. Davis is a postdoctoral fellow in Pulmonary and Critical Care Medicine at Johns Hopkins University. He received his MD from the University of Pennsylvania (2019) after completing an Intramural Research Training Award fellowship at the NIH (2013-2015). He has coauthored 9 peer-reviewed publications and presented 6 oral and 8 poster presentations at national conferences. His department is strongly committed to his scientific career and has awarded him a 1-year fellowship on a T32 training grant. Career Development Plan: In order to achieve his goals, he proposes to develop expertise in biostatistics, clinical epidemiology, and bioinformatics through completing the research aims outlined in this proposal under the close guidance of his mentorship team, as well as through pursuing formal coursework leading to a Master of Health Science (MHS) degree from the Johns Hopkins Bloomberg School of Public Health. Mentors: His primary mentor is Dr. Theodore J. Iwashyna at the Johns Hopkins School of Medicine, who has supervised 15 K-level or equivalent career development awards of whom 10 have so far transitioned to PI of R- level awards or equivalents. He will be co-mentored in bioinformatics by Dr. Khyzer Aziz, assistant professor of pediatrics and Chief Medical Information Officer at the Johns Hopkins Children’ Center, and in biostatistics by Dr. Akihiko Nishimura, assistant professor of biostatistics at the Bloomberg School of Public Health. Aims: He will use the Johns Hopkins Precision Medicine Analytics Platform (PMAP) comprising observational data from 2.4 million patients and 23 million healthcare encounters in both research aims. First, he will perform a target trial emulation evaluating the effect of scheduling early primary care follow-up at discharge following admission for hypoxemic CAP. Then he will develop the informatics infrastructure needed to investigate optimal discharge diuretic regimens for survivors of CAP with comorbid heart failure in his future K23 award proposal. Deliverables from Aims: His proposed aims will lead to 3-4 publications over 2 years with semiannual presentations at national conferences. This proposal constitutes the foundation for a future K23 application to develop evidence-based guidelines to improve outcomes for survivors of pneumonia and respiratory failure by supporting their safe discharge home following hospital admission.

NIH Research Projects · FY 2025 · 2025-08

PROJECT SUMMARY/ABSTRACT Malaria has exerted a selective pressure on the human genome over hundreds of millennia. Evidence suggests that malaria may have also influenced the human microbiome. Animal and human studies previously showed associations between the gut microbiome and malaria severity, the risk of infection, anti-Plasmodium immunity, and antimalarial pharmacokinetics. Conversely, infection with Plasmodium has been shown to acutely alter the gut microbiome, offering for example one explanation for the long-observed association between malaria and non-typhoidal salmonellosis. We propose to investigate various clinical and parasitological relationships between acute P. falciparum infection and the human gut microbiome in the context of a recently completed randomized controlled trial (NCT04009343) and an existing observational cohort, leveraging recent advances in multiomics technology and bioinformatics to pursue mechanistic questions about the malaria-microbiome nexus in a human population. The proposed research is structured according to 3 specific aims that leverage existing fecal specimens and participant data from a recently-completed clinical trial of uncomplicated malaria and forthcoming samples from a Johns Hopkins Malaria Research Institute-funded observational cohort of severe malaria. Aim 1 will deploy metagenomics and metatranscriptomics to characterize the structure and functional potential of the intestinal microbiome and interrogate its relationships to multiple clinical and parasitological outcomes. Aim 2 will leverage existing plasma drug concentration data from a multidose pharmacokinetic design to assess the potential impacts of the microbiome on orally-administered antimalarial drugs. In nonlinear mixed effects models of drug pharmacokinetics, we will test interactions of the primary parameters with structural and derived functional features of the intestinal microbiome. Microbial biochemical pathways of interest that emerge from Aim 1 and 2 results will be tested by integration of metabolomic data generated under Aim 3 with metagenomic and metatranscriptomic data, identifying candidate biomarkers and targets for prognosis, prevention, and adjuvant treatment for future study. The proposed research will generate foundational evidence of clinically and pharmacologically relevant malaria-microbiome interactions in a patient population. More broadly, this work will expand our understanding of the role of the microbiome in disease pathogenesis and drug pharmacokinetics. The investigative team joins malariology and clinical pharmacology expertise with molecular, epidemiological, and computational microbiome-focused expertise to establish a 5-year research project that employs the most up-to-date technology and methods in multiomics to an oft-understudied but globally significant disease. The results would build significantly upon findings from preclinical studies of malaria and the microbiome, which have yet to be translated conclusively to field studies.

NIH Research Projects · FY 2025 · 2025-08

Project Summary/Abstract Pulmonary arterial hypertension (PAH) is characterized by pulmonary vascular remodeling and increased pulmonary vascular resistance. Most therapies for PAH work by reducing pulmonary vascular resistance, without addressing the underlying cellular and molecular mechanisms driving pathobiology. Our prior work in metabolomics has implicated upregulated kynurenine pathway (KP) metabolism as an early, persistent, and potentially causal feature contributing to PAH pathobiology. Our preliminary data suggest upregulation of tryptophan 2,3 dioxygenase (TDO), an enzyme that converts tryptophan to kynurenine, occurs in an expanded fibroblast population in the PAH lung, and increased TDO drives increased kynurenine production in PAH. Kynurenine is an endogenous activator of the transcription factor (TF) AHR (aryl hydrocarbon receptor) in other biologic contexts. Separately, exogenous AHR activation was recently proven necessary and sufficient for PAH development in a preclinical model, and AHR activity is increased in human PAH. However, the downstream effects of upregulated KP metabolism in PAH are undefined, and the endogenous upstream activator(s) of AHR TF activity in human PAH are unknown. We hypothesize that TDO-derived kynurenine is the endogenous activator of AHR in human PAH, and that kynurenine-induced AHR activity in PAH promotes transcriptional programs that contribute to pulmonary vascular remodeling and resistance. Because pharmacologic TDO inhibitors are under study for other applications, interrogation of a TDO-kynurenine-AHR axis may reveal an opportunity for repurposing of a therapy to PAH that addresses fundamental pathobiology. We propose 3 Specific Aims. Aim 1 establishes a prospective observational cohort of PAH patients in which we will measure longitudinal KP metabolites and AHR activity over the treatment course. We will apply a causal inference framework to test AHR activity as the causal mediator of KP metabolite effects on pulmonary vascular resistance and clinical outcomes. In Aim 2, we will perform single-cell RNA sequencing in archived PAH lung specimens to investigate the TDO-expressing fibroblast population using high-resolution clustering, co-expression analyses, and trajectory inference. We will perform cell-cell ligand-receptor signaling and TF activity analysis on high-resolution clusters to identify activators of AHR-mediated transcriptional programs. Our proposed use of archived tissues presents a unique opportunity to analyze KP-AHR cell type-specific transcriptional associations with pulmonary vascular resistance and other patient-level clinical data. In Aim 3, we will perform complementary gain- and loss- of-function experiments in rodent models of PAH to examine effects of TDO forced overexpression and inhibition on KP metabolism and AHR-mediated gene expression in isolated pulmonary vascular cells (e.g., fibroblasts and endothelial cells) in vitro. Further, we will test the molecular and phenotypic effects of TDO inhibition in our animal models in vivo. Should our overall hypothesis be confirmed, the proposed studies will provide a robust evidentiary foundation for early-phase trials investigating KP inhibition as a disease-modifying strategy in PAH.

NIH Research Projects · FY 2025 · 2025-08

Project Summary The goal of this application is to support Michael Sauria, PhD, as a Research Software Engineer (RSE) in the Andersen lab in the Department of Biology at Johns Hopkins University. He currently helps administer the Caenorhabditis Natural Diversity Resource (CaeNDR) under the direction of Erik Andersen, PhD. This project uses natural variation across three species of nematodes and thousands of strains to provide insights and resources for discovery not otherwise possible in genetically homogenous model organisms, such as natural differences in toxicant response and drug resistance in closely related parasitic nematodes. This RSE role will support several funded NIH projects by Andersen and many others by additional investigators across the evolutionary genetics and Caenorhabditis communities. In order to thrive and expand, CaeNDR requires a dedicated RSE, something it has been lacking since the Andersen lab moved from Northwestern University a year ago. This grant would provide Dr. Sauria the dedicated time necessary to oversee this project. In particular, he will overhaul the underlying infrastructure, including rebuilding the analysis pipelines to add robustness, flexibility, and quality assurance as well as migrating the project from Google Cloud Platform to Amazon Web Services. Dr. Sauria will develop additional tools and resources available to CaeNDR users to make better use of the hosted variation data, including strain-specific CRISPR-Cas9 guide RNA design, genome-wide association mapping tools in C. briggsae and C. tropicalis, sister species to C. elegans, and evolutionary statistics. Finally, Dr. Sauria will improve the user experience and user interface of CaeNDR by developing interactive tools for real-time data exploration, guided site and tool walkthroughs, and extensive documentation to support the analysis workflows as their own resource. Ultimately, this grant will support the harmonization of CaeNDR as a benchmark for other model systems, such as yeast, fly, and mouse, so that similar tools can be added to the Alliance of Genome Resources, leveraging natural variation as a powerful discovery platform.

NIH Research Projects · FY 2025 · 2025-08

PROJECT SUMMARY/ABSTRACT The goal of this K24 Mid-Career Investigator award is to provide resources, training and protected time to strengthen and expand Dr. Michelle Eakin's mentoring program that is focused on patient-oriented implementation science. Asthma is the third leading cause for hospitalization in children and the leading cause of school absenteeism in the US. Children from marginalized backgrounds at increased risk for asthma prevalence and morbidity. Despite a strong evidence-base for the efficacy of asthma care programs in reducing asthma morbidity, little progress has been made implementing and sustaining interventions in at–risk communities. Effective interventions require a multi-level approach, and requires engaging community settings such as schools and Head Start programs. These settings are an ideal location for reaching children given the strong connection to families and trust from communities. However, often these low resource settings have limited staff capacity with high rates of burnout which can lower implementation success. Although limited staffing has been noted as a prominent barrier to implementation, there has been relatively little focus or research on how to adapt to overcome this barrier in community settings to improve implementation success. This K24 will leverage ongoing data collection in two NHLBI funded R level projects to evaluate the role of burnout and burden on implementation of asthma programs in school settings. Dr. Eakin is an ideal candidate to provide mentorship in this area based on her expertise in qualitative methods, implementation science and asthma management programs. Dr. Eakin has a strong and growing record of successful mentorship in implementation science across different settings and populations. This K24 award will allow Dr. Eakin to expand her mentoring skills by including trainees from graduate to post graduate training to evaluate burnout and staff burden as a determinant of implementation success and evaluate change in staff wellbeing throughout the implementation process in low resource high stress educational settings. As part of this award, Dr. Eakin will participate in training activities including one and on meetings with mentors, local workshop on master mentoring and structured curriculums on K-to R transition. Dr. Eakin will refine her skills in qualitative methods with advanced training on rapid qualitative approaches. The scientific goal of this project is to investigate the role of staff burnout on implementation success using staff surveys and qualitative methods. Dr. Eakin's scientific aims and training plan will provide an outstanding environment for training of researchers focused on implementation science in a range of settings.

NIH Research Projects · FY 2025 · 2025-08

Parental caregivers of the 1 percent of children in the United States with complex chronic conditions increasingly depend on electronic communication (e-communication) through modes such as patient portal secure-messaging, personal text messaging, and e-mail with the child’s healthcare team in the periods between in-person, face-to-face medical encounters for coordinated, safe management and support. Due to the high-risk and long-term nature of cancer care that requires high-quality communication, pediatric cancer caregivers represent an ideal population for studying e-communication in healthcare. The relational coordination theory, which proposes that optimal care is most effectively carried out through frequent, timely, accurate, problem-solving communication among key project-partners, including caregivers, supported by relationships of shared goals, shared knowledge, and mutual respect, can be applied to healthcare structures to systematically and reliably improve the quality of patient-centered e-communication. While e-communication is becoming widespread, little is known about its quality or impact on care, and evaluation methods are limited. The purpose of this multi-method study is to investigate the quality of e-communication between the caregivers of children with cancer and their healthcare teams by addressing the following specific aims: 1: Compare the frequency and mode of e-communication of caregivers of children aged 0-13 undergoing active cancer treatment with caregiver characteristics to assess differences in use (N=134) 2: Compare surveyed relational coordination scores and satisfaction scores among these caregivers to evaluate the association of e-communication quality and caregiver satisfaction, and further assess whether caregiver characteristics or e-communication mode influence scores (same sample) 3: Explore caregivers’ experiences with e-communication with their child’s healthcare team between episodes of care, focusing on its impact on perceived quality of care (N=20). The proposed study will use quantitative surveys and qualitative caregiver interviews to address knowledge gaps by identifying critical themes of asynchronous e-communication between episodes of in-person medical care. Univariate analysis and multivariate regression modeling will test associations among differences in e-communication utilization, mode, access, relational coordination, satisfaction, literacy, caregiver demographics, and cancer characteristics. Semi-structured qualitative descriptive interviews, guided by the Relational Coordination Theory, will explore the impact of e-communication on perceived quality of care. This research provides a foundation to understand high-risk e-communication, aiming to enhance caregiver-team communication.

NSF Awards · FY 2025 · 2025-08

The Super Dual Auroral Radar Network, or SuperDARN, is an international collaborative experiment for observations of plasma motions in Earth’s upper atmosphere. By observing ionospheric plasma motions, a multitude of geophysical processes are being studied. These processes range from fundamental plasma instabilities to the global-scale plasma response to changes in the solar-terrestrial environment. Each of these areas of study contributes to developing an understanding of the coupling of energy from the Sun into Earth’s upper atmosphere and its effects on humanity and technological systems. This project will support operations and maintenance of the U.S. SuperDARN radars in the northern hemisphere by the consortium of Penn State University, Virginia Tech, Dartmouth College, and the Johns Hopkins University Applied Physics Laboratory. The collaboration operates twelve radars that cover a vast region from Alaska to Iceland at high latitudes, and Oregon to Virginia at middle latitudes. In addition to operation and maintenance activities, the project will support a program of research that exploits new capabilities that have been developed over the last several years. This includes providing improved fidelity in measurements (plasma convection mapping and imaging), extending the area over which measurements are obtained (bistatic observations), and providing new types of measurements (sounding). SuperDARN has a long-standing commitment to including graduate students in all aspects of the program. The SuperDARN observations are also important for space weather applications since HF radio propagation is sensitive to perturbations in the bottomside ionosphere, e.g., solar flares. 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 · 2025-08

This project belongs to the field of dynamical systems, which is the mathematical study of processes that evolve over time according to fixed rules. These processes often exhibit complicated and chaotic behavior yet display underlying patterns that can be described in terms of stationary measures, that is, certain probabilistic objects that remain stable as the underlying dynamics evolve. The focus here is on stationary measures that arise in nonlinear systems, with particular attention to their geometric and analytic structure. The goal is to understand when such measures are absolutely continuous, when their Fourier transform decays, and how these properties relate to the dynamics that generate them. The work will bring together researchers based in the United States and Israel and will involve the training of graduate students and postdoctoral fellows. A particular emphasis will be placed on maintaining strong collaborative ties between research groups working in dynamics, geometry, and analysis. The main technical goal of this project is to study the regularity and dimension of stationary measures arising from nonlinear actions, such as self-conformal systems and random matrix products. When the maps involved are real analytic and satisfy appropriate separation properties, one expects to be able to compute their dimension in simple terms such as entropy and Lyapunov exponents, and to determine whether they are absolutely continuous. The project aims to establish these properties by studying the behavior of the system under repeated iteration and by using tools that reveal how randomness and geometry interact at different scales. The project will develop via methods from hyperbolic dynamics, harmonic analysis, homogeneous dynamics, spectral theory of transfer operators, and additive combinatorics. These include the use of appropriate disintegrations of measures, spectral gaps for transfer operators under appropriate assumptions, and comparisons between different criteria for separation. The broader aim is to clarify how non-linearity results in stationary measures enjoying rich multiscale structures, which in turn governs their analytic and geometric properties, and to use this understanding to characterize rigidity and regularity phenomena in these systems. 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 · 2025-08

Min-max optimization underpins technologies ranging from generative Artificial Intelligence (AI) to large-scale reinforcement learning, yet today’s methods remain slow and unreliable for many real-world tasks. This suboptimality stems from the traditional approach of adapting minimization techniques to the min-max setup, which necessarily overlooks the unique complexities inherent in min-max problems. This project fundamentally revises this approach, developing specialized theoretical frameworks and efficient algorithms tailored explicitly to min-max optimization. By establishing a deeper understanding of these unique characteristics, the proposed research will significantly enhance the efficiency and robustness of min-max optimization, directly impacting practical applications in machine learning and artificial intelligence. Technically, this project will first explore core theoretical foundations under idealized convex-concave conditions, emphasizing accelerated convergence through anchor-type algorithms and enhanced stochastic methods with relaxed assumptions. Building upon this, the project will also develop practical algorithms that are robust to realistic, non-ideal conditions, including methods for nonconvex problems, efficient sampling strategies for stochastic settings, and adaptive update rules. Additionally, the research will investigate efficient alternating-update strategies, proximal gradient-type methods, and applications to training deep neural networks. These efforts are anticipated to greatly enrich the mathematical tools of min-max optimization and to lead to the discovery of more practically efficient algorithms. 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 · 2025-08

Understanding how complex systems evolve over time, from evolving opinions on social networks to the interactions of cells or genes, relies on reconstructing their underlying dynamics from observational data. Many experiments and measurements, however, only provide snapshots of system ensembles rather than complete trajectories, posing a significant challenge for data-driven modeling. This project will develop new mathematical and computational tools to learn the rules governing these high-dimensional systems directly from ensemble data. By doing so, the research will advance scientific discoveries using ensemble data and enable accurate predictions in fields ranging from medical imaging and biological dynamics to opinion dynamics modeling. The methods and software produced will be openly shared, providing hands-on training for undergraduate and graduate students, and establishing an international collaborative network that advances scientific knowledge and trains future researchers. The project will introduce new theoretical and computational tools for learning dynamics from ensemble data. The research uses weak-form partial differential equations and gradient flows to enable learning from the empirical distributions of the ensemble data. It targets both single-system and cross-system learning by developing an automatic kernel regression and a task-specific attentional model. The research will establish mathematical foundations for both single-system and multi-system learning through theoretical analysis of identifiability and convergence rates. The resulting algorithms will be implemented in open-source software, and applied to exemplar problems in imaging and interacting particle systems. This work will advance mathematical theory and provide robust computational tools for data-driven scientific discovery. 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 · 2025-08

The sun routinely produces powerful bursts of energy, such as solar flares and coronal mass ejections, that can travel through space and impact Earth. These solar eruptions are major drivers of space weather, posing serious risks to critical technologies that support everyday life, including GPS systems, power grids, communication satellites, and aviation safety. While there are many space weather forecasting methods available, their long-term reliability and transparency remain major concerns for stakeholders, including scientists, policymakers, engineers, and emergency response planners. Most space weather forecasting methods operate as "black boxes," where predictions are made without clear explanations of how or why certain outcomes are reached. This lack of interpretability reduces trust in the models and limits their practical value in high-stakes decision-making situations. This project addresses this challenge by developing data-driven learning methods that are not only accurate but also explainable. By building these innovative tools to help scientists and decision-makers understand how solar activity connects to space weather events and how predictive models reach their conclusions, the project enhances the accountability, transparency, and usability of space weather forecasts. This project develops modular cyberinfrastructure for interpretable and explainable space weather prediction systems, with a focus on solar transient events. These efforts advance both heliophysics and the field of explainable artificial intelligence. The research integrates cutting-edge methods in interpretable machine learning, uncertainty quantification, and feature attribution to assess how prediction systems perform under diverse, multimodal data sources. Key cyberinfrastructure contributions include (1) Self-interpretable multivariate time series classification models for transient event prediction; (2) Post hoc explainable artificial intelligence methods for image-based and multimodal predictors; and (3) Uncertainty-aware prediction using conformalized techniques to produce reliable, probabilistic forecasts. These components will be delivered through an open-source, reusable framework for prediction, explanation, and visualization, designed for both research and operational use. The infrastructure will enable large-scale experiments with a focus on local/global interpretability and uncertainty quantification in space weather analytics. The project’s three main goals are (1) advancing understanding of solar eruption precursors via interpretable machine learning; (2) contributing new explainable artificial intelligence techniques tailored to high-dimensional, spatiotemporal data; and (3) supporting the complete research-to-operations path and the research-to-operations-to-research loop through trustworthy, actionable forecasting tools. The outcomes of this research will bridge fundamental solar physics and modern data science to deliver both domain-specific insights and reusable tools for science-grade machine learning. This award by the Office of Advanced Cyberinfrastructure is jointly supported by the Division of Atmospheric and Geospace Sciences and the Division of Research, Innovation, Synergies and Education in the Directorate of Geosciences. 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.