New York University

NIH Research Projects · FY 2025 · 2025-09

RESEARCH SUMMARY Goal-directed behavior in dynamic, naturalistic environments depends on a hierarchy of decision-making processes that operate at different timescales to infer context, adjust decision policy, and map stimuli to actions. Still, the neural mechanisms underlying hierarchical decisions are poorly understood. To study these processes, we propose a project based on recordings from large neural populations in the lateral intraparietal area (LIP), supplementary eye field (SEF), and dorsolateral prefrontal cortex (dlPFC) of macaque monkeys performing a hierarchical decision task. This task is a variant of the direction discrimination task with a random dots stimulus in which monkeys classify the motion direction of the stimulus while also tracking a hidden, spontaneously changing environment variable. Preliminary analysis demonstrates that monkeys can perform the task by approximating a normative strategy that requires monkeys to track variables on multiple timescales: the monkey must integrate motion evidence to make individual decisions and must integrate decision feedback and confidence to infer the context over the span of many trials. To understand the representational and computational mechanisms that support these dynamic and temporally multiplexed codes, I hypothesize that hierarchical decision variables are flexibly represented in orthogonal subspaces in each population to enable parallel computations and generalization across conditions. To investigate this, I will study the representational geometry in each recorded region through linear decoding, dimensionality reduction, and manifold capacity analyses. Additionally, preliminary results also demonstrate that the representations of both perceptual decision variables (i.e., dot motion), and contextual decision variables (i.e., environment) are distributed across all recorded regions. I hypothesize that this distributed code is maintained through task-relevant communication subspaces, which I will characterize by identifying representational dependencies and signal latencies across regions. Ultimately, this project will use innovative analyses and a novel, well-controlled task to provide a comprehensive understanding of the neural mechanisms – both within regions and across regions – that underly hierarchical decision-making processes in the primate brain. Clarifying such mechanisms has broad applicability in our understanding of general cognition at large and may ultimately inform new diagnostic tools and therapeutic strategies for conditions that involve abnormalities in decision-making and belief-updating, such as depression, anxiety, and schizophrenia.

NSF Awards · FY 2025 · 2025-09

Non-technical Abstract The aim of this research is to discover the organizational principles of non-equilibrium statistical mechanics. Equilibrium thermodynamics, developed in the 19th and 20th centuries, is used to calculate how machines, animals and plants work. In equilibrium a forward and a backward sequence of events are equally probable, no work is done. Hence: “if you’re in equilibrium you’re dead”. Luckily our world is far from equilibrium. Thermodynamics tells us how energy, particle concentration, etc., are partitioned when systems interact. We will use mathematical models and closely related experimental systems that can be driven continuously from equilibrium to non-equilibrium. As we leave equilibria, we will study how the relationships between, energy and temperature, density and pressure, etc. change, how the equations can be modified. Better understanding of non-equilibrium statistical mechanics will lead to more efficient use and generation of energy, new processing techniques, new materials and better understanding of biology and life. International workforce development is also to be strengthened by the junior researchers working together and learning different techniques and approaches to problem solving. Technical Abstract Many natural and industrial processes take place far from equilibrium where we lack the fundamental organizational principles, the laws of thermodynamics, available in equilibrium. The aim of the research proposed here is to quantify “out-of-equilibriumness”, order, entropy and other measurables and develop their use in describing the properties of dynamic stationary states. The American and Israeli PI’s have developed three new tools: 1) Computable Information Density, CID, an entropy/complexity approximate that indicates ordering, 2) time and length correlations by a CID decimation process, 3) Universal Local Entropy Production, (EP) , by violations of detailed balance comparing the first half of a movie with the time reversed second half. The study involves systems that can be tuned continuously from equilibrium to non-equilibrium while measuring CID, EP as well as the partitioning of particles, pressure, entropy, etc. Studies will include how particles are partitioned when two systems can exchange - is there a new form of chemical potential that holds? A question is whether as one leaves equilibrium the changes can be handled perturbatively. The models that will be studied are interesting because they can be treated theoretically, by simulation and by experiment. These model systems have also led to interesting new dynamical phase transitions and the physical creation of materials that cannot be made from equilibrium. Along with fundamental contributions to science, better understanding of non-equilibrium phenomena will lead to more efficient generation and use of energy, new processing technologies, new materials and new insights into biology and life. 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 · 2025-09

Computational fluid dynamics is an area of engineering that predicts complicated flows, such as air flow around supersonic aircraft, hurricanes, and fuel combustion processes inside an automobile engine. Accurate and rapid prediction of these complicated flows, particularly their chaotic or turbulent patterns which are collectively called nonlinear processes, is an ongoing challenge for engineering. Quantum computers, which rely on the principles of quantum mechanics, can perform calculations at much greater speed than traditional supercomputers. Currently however, quantum computers cannot be used to predict complicated flows. The central problem is to make quantum data processing, which is based on linear processes, work with nonlinear processes associated with complex flows. To address this problem, this project will combine quantum computing with classical supercomputing, using artificial intelligence to accelerate connection between these two computing methods. If successful, this project will enable rapid prediction of complicated flows associated with natural systems such as wind gusts and engineered systems such as supersonic transport. To help disseminate these new methods, the project will host workshops at national scientific meetings, and work with private companies, both large and small, to test out the computer code. Additionally, students will be trained in a collaborative environment that includes engineers, computer scientists, and physicists to build a quantum science and engineering workforce here in the U.S. The lure of quantum computing for solving complex fluid flows is the promise of exponential advantage in memory and speed compared to classical computing. At present, quantum computing is limited to very simple fluid flow models, and the potential for solving complex problems has remained unrealized. Applications of quantum computing for prediction of complex flow phenomena face the inherently nonlinear and dissipative nature of chaotic processes, including turbulence. Novel quantum protocols with end-to-end utility are needed, including data loading, computation, and data readout that accommodate these nonlinear processes. This project will develop new tools for this purpose, including iterative matrix multiplication and inversion, detection of extrema, and approximation of non-unitary and non-Hermitian quantum systems by combinations of unitary quantum bits. The quantum computing component will then be integrated with classical computing and tuned by machine learning to enable prediction of complex fluid flow phenomena. Engineering applications include prediction of supersonic, combustion-driven, and atmospheric flows relevant to national energy and security goals. The proposed research also addresses national needs in quantum computing articulated in the U.S. National Quantum Initiative Act, the National Quantum Initiative Reauthorization Act, and the U.S. CHIPS and Science Act. Anticipated Transformative Impact: Prediction of complex flows e.g. supersonic transport and hurricane formation. 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 · 2025-09

With this award, the Molecular Foundations for Biotechnology (MFB) Program is funding Dr. Oded Regev from New York University (NYU) and Dr. Jef Boeke from NYU Langone Health to develop a novel experimental method (PolySnap-seq) to probe the positions of ribosomes along individual RNA molecules. Ribosomes are molecular machines that produce proteins by “walking” along RNA and reading the instructions written in it. Despite being critical for life, it is currently not clear how ribosomes organize along an RNA molecule during gene expression. Are they equally spaced along the RNA molecule or do they cluster to form “convoys”? Do they maintain a safe distance from each other or pile up in “traffic jams”? The experimental method uses an enzyme to modify the bases of RNAs that are not protected by the ribosomes, thereby allowing researchers to take a snapshot of the simultaneous positions of all ribosomes along an RNA molecule. As such, the method provides novel insights into ribosome collective behavior and enables new biotechnology applications to, for example, optimize gene expression. This project provides graduate students and postdoctoral fellows with specialized training in method development, data analysis, and machine learning. Finally, an outreach component uses “yeast art” to introduce the general public to biotechnology. The efficiency in which ribosomes translate mRNA into proteins is affected by a complex regulatory logic encoded in RNA sequence and structure. Understanding this logic is required for the rational design of RNA transcripts in biotechnology. Powerful tools like ribosome profiling revolutionized our understanding of how ribosomes are affected by RNA features. Yet, many fundamental questions related to translation remain elusive, such as the minimum separation between ribosomes, the extent of burst translation, and the effect of single nucleotide variants. This research project will develop PolySnap-seq, a novel method for probing in bulk the simultaneous positions of all ribosomes along individual transcripts using an adenosine deaminase enzyme and long read sequencing to identify stretches of mRNA that are not occupied by ribosomes and, therefore, can be efficiently modified to inosine bases. The method will provide detailed transcriptome-wide information on translation, including the separation between ribosomes, how such spacing depends on the distance from the translation initiation site, and the effect alleles and isoforms have on ribosome profiles. 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 · 2025-09

This project aims at contributing to the mathematical foundations of risk management in finance and robust generative models. The first major focus is the design of dynamic stochastic models subject to domain and/or distribution constraints. Such models will play a key role for a robust representation of the underlying uncertainties and will allow for a better generation of risk scenarios thus improving the back testing abilities of financial risk management. The second key component addresses model risk and hedging by developing sensitivity analysis tools in the context of distributionally robust optimization. This project contributes new mathematical methods for optimizing systems of interacting agents which plays a crucial role in the analysis of financial risks. More specifically, the project investigates ergodic optimal semimartingale transport problems to model multidimensional stochastic processes under both domain (support) and distributional constraints. These results can be applied to diffusion-based generative methods in artificial intelligence and are expected to outperform standard score-based procedures. A second major part addresses model risk assessment and hedging through the so-called distributional robust optimization. This involves defining model deviations within small Wasserstein balls around a reference martingale model--a novel concept for quantifying and mitigating model risk. When volatility surface calibration gives access to marginals, the project introduces a new notion of static model risk hedging. Extending to continuous time poses a significant mathematical challenge. Finally, the project builds on the primary investigator’s ongoing work in optimal control and differential games involving interacting populations. Key questions include the impact of distributed control in delegation relationship, the structure of optimal stopping strategies in mean field stopping problems, and the effect of optimal debt cross-holding on default propagation in systemic risk models. 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 · 2025-09

Young children begin learning math long before they enter school. Between ages 1 and 3, toddlers develop early understandings of number, shape, size, and location through everyday conversations and interacting with materials in their home environment. But not all children are exposed to the same math talk and objects, and early differences in home experiences may contribute to later gaps in math achievement. This project explores how toddlers' daily interactions with caregivers--including the language they hear and the objects they use--support early math development. It also addresses a longstanding scientific debate concerning the role of language, if any, in shaping cognition. By comparing toddlers whose home languages are either English or Mandarin--two languages that express math ideas in dramatically different ways--this study also offers a rare opportunity to investigate how language structure might influence the formation of early math concepts. Results from this project will lead to a better understanding of sources of individual differences in early math cognition that are foundational for later math achievement. To address key gaps in the understanding of early math learning, this project examines how math-related language and home environments shape toddlers' math-related conceptual development across two linguistically distinct populations. Researchers will video-record English- and Mandarin-speaking toddlers (ages 18 to 30 months) and their caregivers during naturalistic and structured play at home. Fine-grained video analysis will capture the frequency, form, and function of caregiver math language, and the types of activities and objects associated with math talk. Complementary assessments of the physical home environment--using video walkthroughs and 3D renderings--will quantify math-related materials such as books, toys, and puzzles. Toddlers' comprehension and production of math-related vocabulary will be measured through parent report, observational coding, and novel direct assessments. By linking linguistic input, environmental supports, and individual differences in toddler math cognition, this project will yield new insights into early conceptual development and the relation between language and thought. All video data and procedures will be shared through Databrary, a secure research video repository, supporting transparency and enabling broader scientific inquiry in the future. This project is supported by NSF's EDU Core Research (ECR) program. The ECR program emphasizes fundamental STEM education research that generates foundational knowledge in the field. Investments are made in critical areas that are essential, broad and enduring: STEM learning and STEM learning environments, broadening participation in STEM, and STEM workforce development. This project is also supported by the Advancing Informal STEM Learning (AISL) program, which seeks to advance new approaches to, and evidence-based understanding of, the design and development of STEM learning in informal environments. This includes providing everyone multiple pathways for accessing and engaging in STEM learning experiences. 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 · 2025-09

About 12% of U.S. adults have difficulty with mobility, including walking or climbing stairs. Current technologies for helping people manage these issues are often expensive, heavy, and hard to use. The goal of this research is to develop a new paradigm for assistive robotics that will make it possible to take such systems from the laboratory to widely accessible tools for people. The research team will combine advances in robotic exoskeletons for human joints, simulation enabled by artificial intelligence, and other approaches to design lightweight, wearable robotic systems that can be personally controlled. The investigators will also test the systems in different settings to improve their usability. The project will develop new science and technology that have the potential to help people to perform daily activities, along with their quality of life and overall well-being. The project will design and test new assistive robotic systems by integrating artificial intelligence, robotics, biosensors, rehabilitation medicine, gerontology, and neurorehabilitation. The research will use a modular design approach, that can adapt to individual needs and daily activities without extensive calibration. By leveraging computational modeling and physics-informed deep reinforcement learning, the systems will learn control strategies from computer simulations and user feedback to deliver personalized support that addresses an individual’s mobility challenges and needs. The objective is to broaden the reach of robotic mobility support to a much larger population in a wide variety of settings. The resulting enhanced mobility and function can lead to broader benefits, including promoting independent living, employment, and well-being. 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 · 2025-09

This research project focuses on active sequential change-point detection for high-dimensional streaming data under sampling or resource constraints, with numerous important real-world applications, including biosurveillance, environmental monitoring, epidemiology, disaster management, homeland security, quality control in manufacturing engineering, and threat detection. The project aims to develop simple yet effective algorithms that are able to quickly detect undesired anomalies or events, subject to false alarm rates and sampling control constraints, when monitoring large-scale streaming data from complex systems. The results of the research are expected to advance the understanding of real-time anomaly detection and online monitoring of high-dimensional streaming data. Graduate students will also receive training through their involvement in the project's research. This project aims to develop new mathematical, computational, and statistical theories and tools for active sequential change-point detection for high-dimensional streaming data under sampling or resource constraints. Our specific research aims are to develop computationally scalable and statistically efficient algorithms to detect sparse changes in the high-dimensions under two settings of sample control constraints: (i) the sequential design setting where sampling matrices can be sequentially or adaptively chose based on past observed data, and (ii) the random design setting where the sampling matrices are random as in the modern statistics or machine learning literature. Moreover, real-world case studies in anomaly detection or online monitoring will be investigated. Results of the project are expected to significantly advance the state of the art in sequential analysis, change-point detection, multi-armed bandit problems, streaming data analysis, and large-scale inference. 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 · 2025-09

Algorithms permeate our modern world, driving everything from navigation, information storage, and data retrieval. In contrast, biological information is inherently physical, carried by molecules whose shapes determine their interactions with the environment. This NSF-funded program aims to explore and harness the interface between the deoxyribonucleic acid (DNA) “software” and the geometric “wetware” of molecules. The research will begin by developing mathematical tools to distinguish molecules based on their 3D shapes and structures. These tools will then be used to create a new programming framework: “algorithmic shape encoding.” Using small DNA tiles as modular pieces in a molecular-scale 3D jigsaw puzzle, the team will construct increasingly complex structures—drawing inspiration from nature’s ability to link form and function. The expected outcomes include breakthroughs in self-assembling materials, biocomputing, and optical communication systems. In addition to scientific discovery, this program will foster interdisciplinary training across mathematics, engineering, and chemistry from high school to the postdoctoral levels. DNA, with its predictable structure and ability to self-organize at nanometer precision, offers a powerful platform for designing next-generation materials. This project builds on the well-established tensegrity triangle motif to create a diverse set of 3D DNA motifs that self-organize into authentic 3D DNA building blocks. In Aim 1—Unit Design: Encoding Information in 3D DNA Motifs—researchers will identify key structural features of DNA motifs that can encode information through molecular shape. This will involve developing computational tools to predict and constrain topologies and verifying motif structures using X-ray and related techniques. In Aim 2—Algorithmic Shape Encoding for Large 3D Nanomaterials—the focus will shift from individual motifs to overall structural organization. The team will (1) design and characterize quaternary structures with defined chirality, and (2) develop periodic, hierarchical, and fractal-based arrays that require supramolecular-level algorithms rather than sequence-level design. Finally, the project will prototype optical materials capable of light-based computation and readout, paving the way for new advances in nanomaterials and biomimetic 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 · 2025-09

With the support of the Chemical Measurement and Imaging Program in the Division of Chemistry, Professor Alexej Jerschow and his group at New York University will advance the development of Zero- and Ultralow-Field (ZULF) Nuclear Magnetic Resonance (NMR) spectroscopy, a cutting-edge chemical analysis technique with applications in real-time reaction and device monitoring. Unlike traditional NMR spectroscopy, which uses large, expensive magnets, ZULF NMR spectroscopy operates in very low magnetic fields, making it portable and cost-effective, while providing detailed chemical insights. This research will improve the sensitivity and precision of this new measurement modality, enabling its use in applications such as monitoring battery performance, detecting environmental changes, and analyzing chemical reactions in real time. This project will foster international collaborations with researchers in Germany and India, training students in advanced scientific fields, including in spectroscopy, quantum mechanics, and computation, and engaging students through partnerships with Pratt Institute to apply spectroscopy in art conservation. These efforts will enhance global scientific networks, support workforce development in STEM, and make chemical analysis more accessible and portable. The project will focus on developing novel sensitivity enhancement techniques, including indirect detection and optimized magnetic field sweeps, to overcome current limitations in ZULF NMR’s resolution and bandwidth. Researchers will investigate relaxation mechanisms at ultralow fields using advanced computational and experimental methods to improve the method’s sensitivity and versatility. These advancements will be validated through applications in battery electrolyte diagnostics, thereby demonstrating ZULF NMR’s potential for in situ chemical analysis. The project’s interdisciplinary approach, combining NMR spectroscopy, quantum mechanics, and sensor technology, will pave the way for portable, high-resolution chemical analysis tools with transformative potential across chemistry and materials science. 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 · 2025-09

As artificial intelligence (AI) and robotic systems increasingly shape critical decisions in healthcare, infrastructure, and public services, there is growing urgency to ensure that engineers and computer scientists are trained to approach their work with a sense of responsibility. Through an integrated approach that emphasizes the social and institutional dimensions of engineering practice, this project aims to enable technologists to see responsibility along with technical proficiency as central to their role. The project team will develop modular content, full courses, and adaptable frameworks for responsible AI and robotics education. The project will include a large-scale train-the-trainer initiative to prepare current and future faculty to bring responsible AI and robotics materials into their classrooms. In addition to building curricular resources, the project team will conduct a rigorous evaluation to understand how these interventions shape student learning, research culture, and institutional norms. By supporting curriculum reform and sharing resources broadly, this project offers a scalable model for embedding responsibility into the core of technical education. 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 · 2025-09

How do people derive meaning from sentences? In what situations does the language comprehension process break down? Artificial intelligence (AI) language models such as ChatGPT, which appear to understand and use language as proficiently as humans do, might seem poised to provide potential answers to this question—answers that could not only enrich our scientific understanding, but also help address language processing deficits. But for AI systems to fully realize this potential, they need to process language in a similar way to humans. Many distinct lines of research show that this is not the case. One area where the discrepancy between humans and AI is particularly pronounced concerns temporary semantic ambiguity in language: cases where the first few words of the sentence are consistent with multiple interpretations, and only later in the sentence is it clear which of the interpretations is the correct one. Whereas human readers can encounter significant difficulty when they are required to change their interpretation of a sentence, AI models generally do not. The goal of this project is to better understand the reason for this misalignment between humans and AI models, and explore ways of modifying AI architectures to bring them more in line with how humans process language. In this project, the researchers will benchmark success in their model development by comparing how the models process language to how humans process language using a variety of psycholinguistic measurements. By better aligning human and AI language processing, this research will open up new directions to address long-standing limitations of current AI models, such as their need to train on far more data than human language learners do. In more technical terms, this proposal explores the idea that one key difference between human and machine language processing is that humans: (i) entertain only a small number of semantic interpretations of the input at a time; and, (ii) treat incremental semantic inference as a key goal in language comprehension. This is pursued through three interrelated aims. First, the proposed work will explore the unexpectedly positive correlation between a model’s perplexity and its ability to explain human reading times: put plainly, the better the model is at predicting the next word, the less similar its predictability estimates are to those of humans. Second, it explores whether the human-model misalignment can be alleviated by adopting semantic training objectives and leveraging causal intervention techniques to focus the model’s internal representations of semantic context on a small number of possible interpretations of the input. Finally, human experiments will be conducted to test the predictions of the models on novel psycholinguistic stimuli, with the goal of determining if the proposed modifications successfully bring the models more in line with human language processing. 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 · 2025-09

This project advances the mathematical foundations of robust finance and decision-making under uncertain. One major focus is the development of a systematic framework to quantify divergences between stochastic models, with the goal of uncovering the underlying geometric structure of stochastic processes. These advancements will facilitate stability analysis of models used in dynamic decision-making contexts such as finance, climate science, and autonomous systems. A second theme is the exploration of optimal strategies for revealing information to capture and sustain attention --motivated by the question, “What is the most exciting game?” This problem has broad relevance to entertainment, behavioral economics, and education. Graduate students will be actively involved throughout the project. More specifically, the first direction investigates the Kullback-Leibler divergence between martingales. Using tools from convex analysis and optimal transport, the project will establish new functional inequalities, develop numerical schemes, and explore statistical applications of these divergences. The second direction frame information design as an optimization problem involving the Kullback-Leibler divergence. Building on martingale optimal transport theory, the project develops a novel control methodology for transport problems with path-dependent objectives. These results have direct applications in the pricing of financial derivatives -- such as Asian, lookback, and barrier options -- where payoffs depend on the historical path of asset prices. 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 ABSTRACT The proposed study seeks to address the urgent need for theoretically and empirically informed interventions that would address the increasing numbers of unaccompanied minors migrating from rural to urban centers in developing countries for better economic opportunities. This process often results in hazardous child labor defined as work that is mentally, physically, socially or morally dangerous and harmful; interfering with schooling and health and mental health functioning, and leading to several other disproportionate risks. Unaccompanied migrant child laborers’ vulnerability is further intensified by the lack of parental protection and community belonging in the host urban center. The International Labor Organization (ILO) estimates that 9.6% of children (ages 5 to 17) across the globe are child laborers and draws attention to migrant child laborers as an underreported and highly vulnerable group, a significant portion of which are female with no education. Poverty has been identified as the main driver of child labor, with family context also being a critical contributing factor. Sub-Saharan Africa (SSA) has the highest rates of child labor (24%), with Ghana -the focus of this study- registering one of the highest child labor prevalence at 22%, including unaccompanied child migrant laborers. In Ghana, unaccompanied adolescent girls migrate from the Northern region to urban centers in the south to work in the informal economy. Load carrying is the most common type of labor for this population and exposes migrant girls to multiple developmental and health risks. Building on the recently concluded R21 study (with 97 adolescent girls aged 11 to 14 years and their caregivers) that showed high feasibility and acceptability, and promising preliminary impact of the ANZANSI (resilience in Dagbani –local language) combination intervention in the same region, we propose to test its effectiveness in a larger two-arm cluster randomized clinical trial among 960 adolescent girls (age 11 to 14 years) at risk of school dropout nested within 32 public junior high schools in the Northern region of Ghana and their caregivers. The schools will be randomly assigned to one of two study conditions: 1) ANZANSI (FEE+MFG) and 2) bolstered usual care. The intervention will be delivered for 12 months, with assessments conducted at baseline and at 12-, 24-, and 36-month follow-ups post-intervention initiation. The study specific aims are: Aim 1: Examine the short- and medium-term impacts of ANZANSI intervention on the incidence of unaccompanied migration for child labor (primary outcome), and academic progress and psychosocial outcomes (secondary); Aim 2: Examine the impact of the ANZANSI intervention on potential mechanisms of change at the individual, family, and community levels; Aim 3: Evaluate the cost and cost-effectiveness of each intervention condition; and Aim 4: Qualitatively examine participants, facilitators, and school leadership’s experiences with the intervention. Our results will inform approaches to prevent poverty- impacted African female youth’s unaccompanied rural-to-urban migration to engage in child labor.

NSF Awards · FY 2025 · 2025-09

Parent language input is crucial for child language development. Therefore, there has been considerable interest in documenting the kinds of language children hear from their parents and other caregivers. Prior research shows that parents tune their language to children’s abilities, and well-tuned input is important for children’s language comprehension. This project is based on a theoretical framework that predicts that the degree of tuning is affected by similarities and differences between parents and children. The project tests this theory by examining whether autistic and non-autistic parents speak differently when interacting with their children. Studying these differences in tuning informs broader theories about the mechanisms supporting language development. This project also supports translational science by laying the groundwork for helping to improve interventions that support language development in children. This project uses two kinds of methods to study language input to children. The first method involves surveying parents to ask them how they talk with their children, and about what kind of language they believe supports their child's language development. The second method involves recording parents and children as they play together using a protocol designed to elicit a variety of child behaviors, including turn-taking, pretend play, and making requests of the parent. Parent language input during these play sessions is analyzed to measure different language variables, such as the number and types of questions they ask, mean length of utterances, and speaking rate. This project contributes a dataset of parent-child interactions that can be shared with other researchers to accelerate further discoveries about language development. 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 · 2025-09

The project seeks to develop a “miniaturization” of plant factories, using aggregates of plant cells in a small microfluidic device that captures cells and regulates the flow of nutrients and biochemicals like plant growth hormones. The device is expected to mimic the hormonal and mechanical environment, allowing the cells to develop and function as they do in the plant. The concept is similar to the way medical scientists use the organ-on-chip devices that mimic the function of a liver, kidney, or other organs to understand how they work. In a similar way, the plant microfluidic device can be used to test conditions that lead plant cells to produce useful compounds or carry out important functions. For example, the device has potential as an experimental tool that will allow plant biologists to study cell division and regeneration--both critical to developing biotechnology applications. The proposal seeks to improve a prototype version of a microfluidic device by automating plant growth hormone delivery and adding the capability to generate variable gradients, both of which will better mimic the growth and functional environment of cells in the plant. Plants are natural factories that produce the food and medicinal compounds upon which humans rely, including anti-cancer, antimalarial, and other drugs. The proposed development of the device can help in finding cost-effective ways to develop and scale up the production of new natural medicinal compounds. The project will also train New York City public school children in plant biology. New technologies that harness the plant’s ability to produce compounds hold great promise in understanding and utilizing the properties of the plant cell. Going beyond the use of microfluidics to isolate single-cells for gene expression studies, microfluidics can also be used to tightly control and mimic the hormonal, nutritional, and mechanical environment of plant cells and tissues. Thus allowing the study of plant cell processes under stressful or other conditions. Further refinements to the device could ultimately be used to steer plant cells to produce desired compounds in vivo. The project proposes to develop a microfluidic device that controls the chemical environment of isolated plant cells while exerting mechanical forces on cells to mimic conditions in the tissues and whole plant. It builds upon a successful prototype design that is capable of capturing groups of cells in cups, maintaining them for weeks in a culture medium, and allowing for cell wall formation and cell division. This research is translational and will result in new biotechnology. 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 · 2025-09

This project aims to develop a new artificial intelligence system that works alongside mathematicians to tackle problems that have resisted solutions for nearly a century. Recent advances in large language models can generate creative insights and partial reasoning steps, but they often make mistakes and cannot guarantee correctness. In contrast, traditional tools for verifying mathematical proofs offer rigorous guarantees but are not well-suited for automatically navigating the vast search spaces involved in complex mathematical discovery. This research combines the strengths of both approaches: using AI to explore promising ideas and using formal logic to rigorously verify and refine them. As a high-impact test case, the team will focus on the Hadamard Conjecture, a longstanding open problem with applications in quantum error correction, communication systems, and coding theory. The project will also produce open-source tools, educational materials, and outreach programs to broaden participation in advanced mathematics and AI. The research introduces a unified framework with three key components: (1) a self-evolving reasoning pipeline that uses synthetic data to guide exploration of promising matrix constructions; (2) chain-of-thought and curriculum learning to help AI decompose complex mathematical tasks into simpler subproblems, integrate partial solutions, and generalize from simpler to more difficult problems; and (3) formal verification tools, such as Lean, integrated with preference alignment to ensure correctness and enable a self-improving system guided by symbolic proof signals. Together, these elements form a closed-loop system for scalable, trustworthy proof generation. Anticipated outcomes include new Hadamard matrix constructions, practical software for AI-assisted mathematics, and foundational advances in combining learning and logic for mathematical problem solving. 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 · 2025-09

Automatically finding and synthesizing information from rich textual sources can support a wide range of use cases across work, education, and personal use. Artificial intelligence systems are already assisting users to fulfill their information needs, from providing encyclopedic facts to answering complex questions that require multiple steps of reasoning. Despite these triumphs, such systems often provide incorrect or outdated information while sounding plausible and authoritative. Furthermore, compared to conversing with domain experts who can answer our questions, interaction with current systems is limited. Instead of engaging in multi-turn interaction with users, asking clarifying questions or follow-up questions, systems mostly take a passive role, aiming to provide accurate information at once. This project envisions interactive systems that critically reason about textual sources to provide high-quality, up-to-date information. This research will advance how language systems interface with rich knowledge sources: parametric knowledge acquired during the language model (LM)’s massive pretraining, documents prepended at inference time, and users who can provide context for their initial input query. The devised systems will model the complexities of real-world scenarios, where users' questions are ambiguous, answers continuously change based on the context of the interaction, and heterogeneous knowledge sources contain imperfect and outdated information. It will develop both data-centric and algorithmic approaches to achieve such goals, (1) expanding the definition of document relevancy to incorporate extra-linguistic contexts, (2) constructing synthetic data to update parametric knowledge and instill multi-document reasoning ability, and (3) developing algorithms to leverage simulated multi-turn interactions. Together, the research will improve how information seeking users interact with systems and how systems interact with knowledge sources. It will enable building systems for wider domains where single-turn interaction over a clean knowledgebase is not feasible. 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 Prior research has demonstrated that financial strain arising from criminal justice involvement is associated with food and housing insecurity, increased recidivism, and overdose deaths; however, there is a paucity of evidence informing approaches to avert these outcomes among people with substance use and criminal justice involvement. This study will address this critical knowledge gap by generating causal evidence of the impacts of legislation (House Bill 139; HB-139) eliminating court fees in New Mexico in 2024. Utilizing the policy as a natural experiment, we will compare drug court engagement and state-wide overdose mortality between New Mexico and a synthetic control group to isolate the impacts of removing court fees and surcharges on these outcomes. Through Aim 1, we will conduct scientific legal mapping to characterize state-level court fee policies nationally (1a), generating a sampling frame to identify plausible counterfactual settings against which to compare overdose deaths in New Mexico before (2022-2024) and (2024-2026) after the passage of HB-139. We will then generate a synthetic control state using data from that sampling frame and conduct controlled interrupted time series analyses to estimate the specific impact of the policy intervention on overdose mortality in New Mexico, using CDC fatal overdose data (1b). Through Aim 2, we will utilize data from New Mexico’s unified court system and conduct interrupted time series analysis measuring the impact of this legislation on court debt, drug court enrollment and completion and recidivism among people in the New Mexico criminal justice system (2022-2026). Through Aim 3, we will conduct qualitative in-depth interviews with stakeholders across New Mexico’s policy and criminal justice systems who were involved in developing, passing, and implementing this legislation (N=20, Group 1), and intended policy beneficiaries, i.e., individuals with court debt and intersecting substance use issues (N=15, Group 2), to characterize the scope and implementation of the legislation. We will utilize the EPIS (Exploration, Preparation, Implementation, Sustainment) Implementation Framework to explore inner and outer contextual factors enabling policy adoption, barriers and facilitators to implementation, and effects on drug court participation, court debt burden, and other issues relevant to defendants with substance use. This study is responsive to RFA DA 25 062 through advancement of scientific understanding and production of actionable policy recommendations to address a known driver of cyclical criminal justice involvement and overdose. Findings will provide insights regarding the burden of court fees on NIDA populations of interest and offer causal evidence regarding their elimination on substance use outcomes. In addition, this study’s focus on policy implementation can provide NIDA and state policymakers with guidance about the adaptation or implementation of similar strategies elsewhere. Taken together, study findings will provide critical evidence to inform strategies to mitigate the impacts of criminal justice involvement on individuals experiencing substance use issues in the US. This study is part of the NIH’s Helping to End Addiction Long-term (HEAL) initiative to speed scientific solutions to the national opioid public health crisis. The NIH HEAL Initiative bolsters research across NIH to improve treatment for opioid misuse and addiction.

NIH Research Projects · FY 2025 · 2025-08

Project Summary Fungal infections are a pressing and underappreciated threat to global public health, causing an estimated 2.5 million deaths each year. One major contributing agent is Cryptococcus neoformans, a globally distributed opportunistic fungal pathogen capable of causing life-threatening meningitis in immunocompromised individuals. Available treatment options for cryptococcal infections and fungal infections in general are limited, underscoring the urgent need for new antifungal development. Two main challenges in combating fungal pathogens are the cellular similarities between fungi and human hosts and their abilities to evade the host immune system. In the case of C. neoformans, the typical small haploid yeast cells undergo a striking morphological transition to form giant, highly polyploid “titan cells” to escape phagocytosis by host immune cells. A key to C. neoformans’s ability to proliferate and form titan cells in the host is the cell cycle machinery. Despite being essential for pathogen proliferation and pathogenicity, the cell cycle machinery has been traditionally excluded as a target for antifungal drug development due to its high conservation between fungi and humans. However, there are differences and modifications in the cell cycle program that are fungal-specific or unique to C. neoformans. Here, I propose to exploit these cell cycle variations essential for fungal proliferation or immune evasion for antifungal drug development. To realize this vision, I have identified two promising cell cycle variations to investigate in C. neoformans. Aim 1 will focus on a fungal-specific cell cycle pathway known as the mitotic exit network (MEN). The MEN is highly conserved in fungi and functions in a fungal-specific cell cycle checkpoint in the budding yeast Saccharomyces cerevisiae, thus providing a potential target for broad-spectrum antifungals. This proposal will determine the function and signaling mechanisms for the MEN in C. neoformans. In addition, we will develop a strategy to perform high throughput genetic interaction mapping using CRISPR/Cas9-mediated gene editing and an inducible degron system to dissect the regulatory network surrounding the MEN in C. neoformans. The goal of Aim 2 is to pinpoint the specific cell cycle alterations that program the reversible titan cell formation. In addition to promoting immune evasion via polyploidization, the titan cells undergo an unusual reductive division to produce haploid daughters to facilitate dissemination to distant tissues such as the brain. We will systematically track and perturb cell cycle progression during titan cell formation and division to distinguish potential models for the polyploidization and reductive division. In parallel, we will perform high throughput phenotyping with a genome-wide depletion library to identify key regulators for the cell cycle alternations in both processes. The proposed study will help identify new drug targets for the important human fungal pathogen C. neoformans and provide a proof of principle for the unconventional and generalizable approach of targeting the cell cycle for antifungal drug development.

NIH Research Projects · FY 2025 · 2025-08

PROJECT SUMMARY Oral cancer patients suffer severe pain which intensifies in metastatic disease. Opioid-based pain manage- ment is inadequate, and marred by the onset of drug tolerance and addiction. There is an unmet clinical need for effective non-opioid analgesics to treat oral cancer pain. Oral cancer derived pain mediators activate and sensitize transient receptor potential vanilloid 1 (TRPV1) channels on trigeminal nociceptors (pain sensing neu- rons). The TRPV1 channel has emerged as a key regulator of oral cancer pain. Transcriptional profiling re- vealed 40 “pain and metastasis genes” overexpressed in metastatic cancers from patients reporting high pain compared to non-metastatic cancers. Fibronectin (protein FN1) is the most highly overexpressed pain and me- tastasis gene in oral cancer. An alternatively spliced variant of FN1 containing the “extra” type III module, EDA (Fn EDA) is only expressed in pathological conditions such as diabetes and cancer. Fn EDA is overexpressed in oral cancer and promotes tumor cell migration and invasion. There is a gap in our knowledge regarding the role of FN1 and Fn EDA as pain mediators. While studying a potential role for FN1/Fn EDA in oral cancer pain we identified that Fn EDA increased the excitability of trigeminal ganglion neurons, supporting the hypothesis that Fn EDA mediates oral cancer pain via sensitization of TRPV1+ trigeminal nociceptors. The long term goal is to identify key targets in the oral cancer microenvironment which can be utilized to alleviate oral cancer pain independent of the opioid pathway, thereby improving quality of life for patients with oral cancer. The overall objectives for this application are to (i) elucidate the clinical FN1 and Fn EDA profile in metastasis-associated pain, (ii) evaluate Fn EDA function within the nociceptive cascade, and (iii) investigate TRPV1 sensitization by Fn EDA. The central hypothesis is the identified pain and metastasis genes FN1 and Fn EDA are pain media- tors, which was formulated from study of oral cancer patients and samples, cell lines and orthotopic cancer- pain mouse models. The rationale for this project is identification of specific mechanisms of receptor activation on nociceptors in the oral cancer microenvironment will provide the foundation to develop non-opioid analge- sics to treat oral cancer pain. Three aims are proposed to test the hypothesis. Aim 1 will evaluate FN1 and Fn EDA expression in oral cancers, and analyze correlation with metastasis-associated pain using patient tissues, clinical pain scores, and multiplex immunohistochemistry. Aim 2 will evaluate Fn EDA as a nociceptive media- tor by manipulation of Fn EDA (overexpression/inhibition), and evaluate effects on nociception using mouse models of oral cancer pain. Aim 3 will evaluate the Fn EDA-TRPV1 pain axis using electrophysiology and cal- cium signaling. The proposed research is innovative as it initiates a new line of research evaluating Fn EDA as an oral cancer pain mediator. The proposed study is significant, because mitigation of oral cancer pain via Fn EDA blockade will establish a paradigm for Fn EDA inhibition in pain management strategies for other cancers that may diminish or eliminate reliance on substance use disorder-prone opioids.

NSF Awards · FY 2025 · 2025-08

A key method to learn about the universe is to use the spatial location of galaxies as distributed in the sky, which are not random but rather clustered due to gravity. These studies allow us to probe critical cosmological questions, such as what were the conditions in the early universe, how much matter is in the universe, and why its expansion is accelerating. The main challenge in extracting such information from astronomical data is the need to precisely compute the effects of gravitational clustering beyond the largest scales. In this project, a team from New York University will develop a new theoretical approach that can incorporate the effects of crossing of the orbits of particles that necessarily happens on small scales, and compute its effect on the observables used to extract cosmological information. This promises to increase the power of galaxy clustering measurements from astronomical data. The project team will integrate aspects of this research into undergraduate and graduate courses and will develop and new cosmology lecture series for the general public. This proposal focuses on extending the formalism of Vlasov Perturbation Theory (VPT), which describes gravitational clustering in cosmology incorporating the effects of orbit crossing that source a velocity dispersion tensor. The major goals of this proposal are to (1) extend the current results of VPT for the matter power spectrum to the matter bispectrum at one-loop order, (2) extend the VPT approach to perturbative redshift-space distortions predictions for the one-loop matter bispectrum, (3) incorporate VPT predictions for matter correlators to predict the one-loop galaxy power spectrum and bispectrum in real space, (4) extend the velocity-difference generating function (VDG) approach to redshift-space distortions to VPT, and (5) build an emulator of VPT to substantially improve the speed of calculations of the galaxy power spectrum and bispectrum in redshift space using the VDG approach for biased tracers. The team will update the publicly available COMET code to use VPT to compute the one-loop galaxy power spectrum and bispectrum in redshift space given a cosmological model that can be used for cosmological parameter estimation through Markov-Chain Monte Carlo. This will be useful to analyze current galaxy redshift surveys such as DESI and Euclid. 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 research will advance the progress of science by creating a set of new statistical tools for analyzing complex networks that are fundamental to the nation's prosperity and welfare. Understanding the underlying structures of these networks is critical for making informed, data-driven decisions to promote better and higher productivity in academia. This project will analyze the complex networks of faculty hiring between U.S. universities to understand how factors such as institutional research productivity, geography, and field of study influence hiring dynamics. The outcomes will enhance the efficiency of the U.S. academic system and provide valuable insights for researchers and policymakers. A key guiding principle of this project is a commitment to broad engagement; all outreach, recruitment, and participatory activities are designed to be fully open to all Americans. The project will also create a faculty hiring dataset with open access to the public, release all new methods in a free software package, and develop training opportunities for the next generation of American data scientists. From a technical perspective, this research will create a versatile statistical toolkit for analyzing weighted, directed networks, which pose significant challenges for existing methods. The investigators will develop four novel methodologies designed for commonly seen applications in analyzing the hiring networks. First, the project will establish a network-to-covariate regression model to handle count-based network data while accounting for complex dependencies between connections. Second, the research will introduce a nonparametric testing framework using network U-statistics to rigorously test for dependence structures. Third, a new method will be developed to identify and perform inference on "core-periphery" structures, allowing researchers to distinguish informative patterns from non-informative ones. Finally, the project will introduce a conformal inference framework to formally compare entire populations of networks, even when the networks differ in size. These new statistical methods will be validated through simulation and applied to the comprehensive faculty hiring network dataset, with results disseminated through peer-reviewed publications and the project's open-source software. 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

Global sea levels have fluctuated dramatically in the past and are expected to continue doing so in the future. It is crucial to be able to forecast future global sea level as a significant portion of the world’s population and built environment is located along vulnerable coastlines. The most crucial uncertainty in predicting future sea level rests with the behavior of the great ice sheets, particularly the marine-based Thwaites Glacier in West Antarctica. As glaciers melt and calve into the ocean, they contribute to a rise in the global sea level. It is necessary to collect all existing observational data in the ocean surrounding the Thwaites Glacier and to put that data into a mathematical equation, allowing a glacier forecast to be created. This is analogous to a weather forecast typically found on a smartphone. The estimates would provide society with a prediction of the global sea level change over this current century and beyond. This project will incorporate recently collected observational data in and around the Thwaites Glacier into a computational ocean model with a dynamically static, thermodynamically interactive representation of the ice shelf. This project is a follow-up to the International Thwaites Glacier Collaboration project, which focused on observing and modeling the Thwaites Glacier system. There is an urgent need to integrate the collected data into a model, particularly at the grounding zone of the glacier, while the necessary familiarity and expertise to understand the data and incorporate it into a modeling framework are available to the community. This data-model assimilation holds strong promise for delivering a deeper insight into the physical mechanisms that guide the evolution of the Thwaites system and its potential to impact the rise of global sea levels significantly. We will use the MITgcm ocean model to simulate ocean circulation within the ocean cavity beneath the glacier. A high-resolution mesh will enable increased resolution in critical regions, such as the grounding zone and areas with complex coastal geometry. The high-resolution mesh will also accurately represent the geometry of the seaside, seabed, and ice base. The model developed for this project will be the most detailed model ever made of the ocean cavity and potentially represent a significant step forward for modeling ocean–ice interaction in general. 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 · 2025-08

Language makes it possible to communicate complex thoughts by composing together simpler elements. How does the brain accomplish this process? AI language models based on transformers, such as OpenAI’s GPT-4, appear to provide a potential answer to this question: like the brain, their behavior emerges from a complex interaction of a massive number of simple units, and they can learn from experience to process language surprisingly well. Yet they also differ from the brain in crucial ways; in particular, their working memory capacity is vastly greater than that of humans. Because they operate under very different constraints than humans, it is unlikely that their representations will match the ones used by humans, and as such they do not immediately constitute good candidates for models of the brain. This project will address this gap, following two main aims. (1) We will develop language models with working memory constraints that better approximate those of humans, and compare their behavior to that of humans in memory retrieval and language processing experiments. We will also test the hypothesis that memory constraints facilitate the emergence of structure by comparing the models’ language acquisition trajectory and efficiency to children’s. (2) We will conduct experiments with human participants, using fMRI and intracranial recordings, as they listen to or read sentences that require maintaining dependencies between words across multiple other words, taxing working memory. The sentences will either be embedded in a story, or presented in isolation (enabling greater experimental control). We will determine to what extent the models we developed in Aim 1 are able to explain these new neural data, as well as existing data from EEG, MEG and fMRI experiments. Overall, this project will develop neuroscientifically plausible AI language models constrained by data recorded from the human brain. Such models can then be used to deepen our understanding of language in the brain, which could advance treatments for language impairments. As such, the project is closely aligned with NIBIB’s mission to develop biomedical technologies that integrate engineering with the physical and life sciences to solve complex problems and improve health. RELEVANCE (See instructions): AI technology has a tremendous potential to help understand how the brain processes language and to address language processing deficits, but to realize this potential, we need AI systems that are similar to humans. We plan to create AIs that, unlike most standard AIs, have working memory that mimics that of humans. The development of this technology will be tightly integrated with existing data from human neuroscience as well as data from neuroimaging and intracranial recording collected for this project.