49082 research outputs found
Sort by
Identification and characterization of a lipopolysaccharide binding, anti-inflammatory protein from Parabacteroides distasonis
No full text2/28/2026 12:00:00 AM
Assessing the impact of institutional and sociodemographic factors on treatment and outcomes in head and neck cancer
Get PDF2026Disparities in head and neck cancer (HNC) outcomes reflect the complex interplay of social, institutional, and biological factors that influence access to care, treatment delivery, and survival. This dissertation examines structural and clinical determinants of inequities in cancer care through three population-based studies using data from the National Cancer Database (NCDB). Collectively, these studies evaluate the effects of healthcare system characteristics, global crises, and evolving treatment paradigms on outcomes among HNC patients. The first study assessed the association between hospital safety-net burden and survival among patients with head, neck, and thyroid cancers. Although crude analyses showed worse survival in hospitals with higher safety-net burden, the association was largely explained by patient and treatment characteristics. Using propensity score–based methods, safety-net burden was not independently associated with overall survival, indicating that differences in outcomes reflect patient mix and access-related disparities rather than hospital performance itself.
The second study evaluated the impact of the COVID-19 pandemic on diagnosis, treatment, and care delivery for HNC patients in the United States. The pandemic year (2020) showed a 17% relative decline in new cancer diagnoses. Among treated patients, there was a small increase in advanced-stage presentation and a shorter time from diagnosis to treatment initiation, while treatment utilization (surgery, radiation, chemotherapy) and radiation discontinuation were essentially unchanged compared with the pre-pandemic period. While baseline disparities by race, income, and hospital type persisted, the pandemic did not substantially exacerbate these inequities.
The third study assessed reduced-dose radiation (50–<66 Gy) versus standard dose (66–70 Gy) for HPV-positive oropharyngeal cancer. After inverse probability weighting and quantitative bias analysis for smoking, overall survival was comparable across dose groups. Survival patterns varied by treatment modality: reduced dose appeared beneficial in surgical/adjuvant settings and less favorable in definitive chemoradiation. No survival disparities were detected by race or income.
Together, these studies provide evidence that institutional context, systemic disruptions, and treatment innovations intersect to shape cancer care outcomes. The findings emphasize the need for equity-centered policies that strengthen safety-net infrastructure, ensure resilience during public health crises, and judicious implementation of emerging treatment innovations such as de-escalation therapies within an appropriate clinical context
The value of school environmental monitoring in a time of change
No full text2026Each day, ~50 million children in the U.S. attend K–12 public schools during a critical period of development. Optimal classroom environmental conditions support student well-being and learning. However, aging school buildings, outdated mechanical systems, limited school indoor environmental policies, and constrained budgets are slowing improvements in classroom indoor environmental quality (IEQ). These challenges can be exacerbated by climate change and have potential lasting implications for students’ future opportunities. Real-time classroom IEQ data can help schools identify environmental problems, prioritize upgrades, advocate for funding, and inform heat preparedness plans. However, the vast and complex nature of IEQ data may prevent stakeholders from unlocking its full potential.This dissertation employed a mixed-methods approach to expanding the value of IEQ monitoring in schools with a climate lens, in collaboration with a large public school district in the U.S.
First, we characterized school thermal variability by air conditioning (AC) and floor level using traditional and novel heat metrics, analyzing minute-level temperature data from a first-of-its-kind network of thousands of monitors. Our findings emphasize the critical need for mechanical cooling in classrooms and exemplify how IEQ data can inform school heat preparedness plans.
Next, we quantified the effects of floor level, AC status, and roof albedo on classroom temperature using Bayesian multilevel models, and estimated hypothetical changes in test scores resulting from climate adaptations. Our findings suggest that central cooling systems are more effective at maintaining comfortable indoor conditions and highlight the learning benefits of moving students to lower-level classrooms, installing light-colored roofs, and upgrading AC systems.
Lastly, we conducted interviews with staff members in school districts pioneering IEQ monitoring to identify barriers and motivators for leveraging this data. We learned how this data supports their work and identified limited environmental health and data literacy as obstacles to fully capitalizing on this wealth of information. Suggested solutions to bridge this knowledge gap included artificial intelligence (AI) tools and partnerships between schools and academic researchers.
Together, these findings highlight the transformative potential of classroom IEQ monitoring to expose disparities in indoor environmental conditions, enhance student well-being and learning, and inform school climate resilience plans.2028-01-14T00:00:00
The ICU resilience project: establishing occupational therapy practitioners' role in addressing patient mental health in the intensive care unit
Get PDF2025Patients in the intensive care unit (ICU) live through harrowing life experiences that can lead to lasting physical, cognitive, and psychosocial changes impacting survivors' quality of life called Post-Intensive Care Syndrome (PICS). Research on strategies to address mental health concerns such as depression, anxiety, and post-traumatic stress disorder within early rehabilitation is still emerging. Occupational therapy practitioners (OTPs) feel most comfortable addressing physical and cognitive impairments in the ICU despite the profession's foundational roots in mental health treatment. This doctoral project aims to address OTPs' limited knowledge and training of their role in addressing mental health concerns in the ICU by developing an evidence-based multimedia website entitled The ICU Resilience Project: Your Occupational Therapy Resource and Toolkit. This website aims to enhance clinicians' understanding of the prevalence of mental health issues following ICU admission and their impact on survivors' quality of life. It seeks to expand knowledge of screening tools, assessments, and treatment strategies within the scope of occupational therapy practice, ultimately boosting the confidence of OTPs in effectively implementing these intervention approaches into daily practice. By embracing this role in the ICU, the hope is for OTPs to assist in decreasing the prevalence of lasting mental health conditions experienced by ICU survivors and increase overall ICU survivor’s quality of life.Keywords: occupational therapy, intensive care unit, mental health intervention, post-intensive care syndrome, psychosocial treatmen
Economic frameworks for spectrum coexistence in advanced wireless networks
No full text2025Electromagnetic spectrum is a renewable shared resource featuring a diverse array of agents attempting to facilitate communications on varying frequency bands. The ever-increasingly interconnected world has resulted in demand for spectrum utilization ranging from Mobile Edge Computing (MEC) clusters supporting smart city applications, to expansions of commercial cellular telephony and broadband internet services in environments where fiber optic deployments are impractical or temporarily disrupted, to support of highly localized industrial automation networks. The advent of the Fifth Generation wireless networking standards has enabled such demands to be realized. However, pre-existing spectrum allocations, especially in the most desirable blocks of spectrum along the mid-band (for propagation balance) and Terahertz bands (for transmission bandwidth), complicate deployments of networks. While in many cases these existing allocations are protected for Public Interest reasons, including scientific, defense, or information dissemination purposes, incumbent frequency utilization is not continuous in nature. This characteristic raises questions regarding allocations of resources to agents within wireless networks. In particular, we propose the use of Open Access Spectrum allocations, whereby commercial entities pay-per-request for access to spectrum not being used by Public Interest incumbents. This leads to two immediate questions: is such a framework viable for its participants, and what is the impact of segmenting commercial users into multiple tiers? The latter question arises from the notion that offering a general authorized use provision may be necessary to incentivize commercial utilization. In addition however, we must consider what occurs with specific applications utilizing spectrum. In particular, MEC networks are dependent not just on spectrum access but on cloud computing access, and are thus vulnerable to Economic Denial of Sustainability (EDoS) attacks generating large amounts of compute resources. This leads to a third question: how do cloud provider actions impact the MEC's susceptibility to EDoS? Utilizing game theory to formally define agents, preferences, and actions, we propose three thrusts of investigation to answer these questions. In Thrust (I) we analyze interactions between Earth Exploration Satellite Service-passive radiometers and commercial users under the Open Access framework and a join-or-balk scenario, and determine that equilibria are unique and socially optimal even under a revenue maximizing objective. In Thrust (II), we consider a scenario where customers have the option to join a paid primary or general authorized secondary tier. Here, maximum revenues will be guaranteed but corresponding equilibria are not necessarily unique nor is the revenue maximizing state necessarily socially optimal. In Thrust (III) we consider behavior in a scalable MEC cluster connected to a network of Internet of Things devices. We find that billing models that enforce minimum charges on clusters increases the incentive to launch EDoS attacks compared to models that do not enforce such minimums. Additionally, randomizing scaling is shown to have potential to disrupt EDoS attack effectiveness while minimizing impacts on legitimate traffic
Maine Home Health United: an online community of practice to support home health occupational therapy practitioners in Maine
Get PDF2025Home health services are medical services provided in the home by a skilled licensed provider (Rural Health Research Center, 2017). Evidence shows that health professionals in rural home health settings experience high levels of stress, isolation, and burnout (Manson et al., 2020). Occupational therapy practitioners (OTPs) are critical members of the home health care team (Roots and Li, 2013). In the state of Maine, an overall aging workforce and patient population, coupled with the COVID-19 pandemic and recent vaccine mandates, have led to severe staffing shortages in all aspects of healthcare. Due to these factors, home health OT practice in Maine has proven to be challenging and ever changing. As a result, Maine home health OTPs often lack knowledge and confidence when providing care to their patients in the homecare settings. Currently, there are no formal support systems for Maine home health OTPs to gain needed knowledge and provide guidance for these issues. As a result, Maine home health OTPs often report increased episodes of professional burnout and poor job satisfaction.
Communities of practice (COPs) have been shown to facilitate peer communication and knowledge sharing, reduce isolation, and increase intention to work in rural areas (Bikinesi et al., 2020, p. 2). The Maine Home Health United program aims to connect Maine home health OTPs through an online COP. It will consist of two components: 1) a Project ECHO® (Extension for Community Healthcare Outcomes) education series using the Zoom platform, and 2) the establishment of a WhatsApp multimedia group. Engaging in Maine Home Health United has the potential to increase connectivity, social support, and feelings of belonging among members by learning and sharing knowledge. This connection has the potential to improve overall confidence with service delivery and in turn, decrease rates of professional burnout and job satisfaction amongst Maine home health OTPs
LSD1- induced signaling mechanisms inhibition sensitizes oral cancer for chemotherapy and immunotherapy
Get PDF2025INTRODUCTION: Lysine-specific demethylase 1 (LSD1) is a nuclear histone demethylase. Our work shows that LSD1 expression progressively increases with tumor grade and stage in clinical oral squamous cell carcinoma (OSCC). The goal is to evaluate the LSD1 mechanism in OSCC by proteomics analysis and its application for immunotherapy and chemotherapy.METHODS: The animal experiments were approved by the Institutional Animal Care and Use Committee, Boston University. The conditional LSD1 knockout mice (K14Cre-Lsd1; Lsd1-/-) and littermates (Lsd1WT/WT) were treated with 4-Nitroquinoline 1-oxide (4NQO) followed by histology and proteomics analysis. Next, 4NQO-treated mice for 20 weeks were subjected to LSD1 inhibitor (SP2509) alone or in combination with Hippo signaling regulator YAP inhibitor (Verteporfin), anti-PD-1 or anti-PD-L1 antibodies for five weeks, followed by IHC and mRNA expression analysis.
RESULTS: Global unbiased proteomics analysis of 4NQO-induced OSCC from LSD1-/- compared to LSD1WT/WT (n=7/condition) showed inhibition of 40 differentially regulated proteins, reduction of canonical pathways signaling including EIF2, mTOR, VEGF, Integrin, ERK/MAPK, Glucocorticoid Receptor Signaling, Acute Phase Response and oxidative phosphorylation evaluated by IPA analysis. LSD1 inhibition attenuates EGF-, YAP-, and pro-inflammatory cytokine-induced pathways involved in oral cancer cell proliferation, migration, and oncogenic transformation. These pathways are validated in HSC-3 and CAL27, and LSD1 knockout OSCC mice. Next, studies in OSCC mice showed that the topical application of LSD1 inhibitor in combination with YAP inhibitor provides sensitivity to OSCC as compared to a single drug alone. LSD1 and YAP co-regulate each other. Finally, genetic, or pharmacological LSD1 inhibition showed to promote PD-L1 expression, thereby sensitizing to the combination of LSD1 inhibitor to anti-PD-1 therapy. Thus, LSD1 inhibition sensitizes to chemotherapy and immunotherapy combinations.
CONCLUSION: We showed for the first time that blocking LSD1 inhibits OSCC, a feed-forward loop oncogenic protein that exists in OSCC and promotes anti-tumor immunity. Next, the detailed mechanism of LSD1 in vitro and in vivo is being evaluated for application in chemotherapy and immunotherapy
On the deep-water and shallow-water limits of the intermediate long wave equation from a statistical viewpoint
Get PDFWe study convergence problems for the intermediate long wave (ILW) equation, with the depth parameter > 0, in the deep‐water limit ( → 0) and the shallow‐water limit ( → 0) from a statistical point of view. In particular, we establish convergence of invariant Gibbs dynamics for ILW in both the deep‐water and shallow‐water limits. For this purpose, we first construct the Gibbs measures for ILW, 0 < < ∞. As they are supported on distributions, a renormalization is required. With the Wick renormalization, we carry out the construction of the Gibbs measures for ILW. We then prove that the Gibbs measures for ILW converge in total variation to that for the Benjamin–Ono (BO) equation in the deep‐water limit (). In the shallow‐water regime, after applying a scaling transformation, we prove that, as → 0, the Gibbs measures for the scaled ILW converge weakly to that for the Korteweg–de Vries (KdV) equation. We point out that this second result is of particular interest because the Gibbs measures for the scaled ILW and KdV are mutually singular (whereas the Gibbs measures for ILW and BO are equivalent). In terms of dynamics, we use a compactness argument to construct invariant Gibbs dynamics for ILW (without uniqueness). Furthermore, we show that, by extracting a sequence , this invariant Gibbs dynamics for ILW converges to that for BO in the deep‐water limit () and to that for KdV (after the scaling) in the shallow‐water limit (), respectively. Finally, we point out that our results also apply to the generalized ILW equation in the defocusing case, converging to the generalized BO in the deep‐water limit and to the generalized KdV in the shallow‐water limit. In the non‐defocusing case, however, our results cannot be extended to a nonlinearity with a higher power due to the nonnormalizability of the corresponding Gibbs measures.Accepted manuscrip
Crystal ribcage: development of a platform to probe real-time lung function with cellular resolution in health and disease
Get PDF2025The lungs are always in motion and exposed to chemical, mechanical, biological, and immunological stressors creating pathologies at cell scales that effect its functioning. Clinical imaging of the functioning lung includes computed tomography (CT) scans and magnetic resonance imaging (MRI) both of which lack the spatial resolution to image cell- scale dynamics. Histology of the lung provides immense cellular detail, but it is only a snapshot in time lacking all dynamic information. With advances in intravital microscopy, circulatory dynamics of capillaries around alveoli (the functional unit of the lung) have been imaged, however the local respiratory motion was lost as the lung was adhered to the imaging surface by glue or vacuum. The goal of this dissertation was to advance a crystal ribcage — an optically transparent, biocompatible, and physiologically informed housing for a functioning ex vivo lung that can be used for real-time cell-scale optical microscopy. The crystal ribcage was first developed for mouse lungs and used to study altered lung mechanics in metastatic breast cancer, pneumonia, emphysema, and pulmonary fibrosis. Specifically in cancer we found (i) nodular tumors grow into alveoli to limit function while co-optive tumors do not alter function; (ii) the extracellular matrix was highly remodeled in nodular vs co-optive metastatic tumors; (iii) the effect of breathing on single cancer cells trapped in capillaries; and (iv) the critical nodule size that suppresses alveolar function. We scaled up the crystal ribcage from mouse to pig and engineered multiple solutions including reinforcement of the ribcage with slender, rib-like structures, development of a synthetic diaphragm, and finally engineering an active abdomen to actuate the pig lung as close to physiologically as possible. This development pipeline was deployed to fabricate non- human primate and human crystal ribcages as a proof of concept. In summary, mouse, pig, non-human primate and human crystal ribcages were developed by overcoming several engineering challenges. The mouse model was used to study multiple models of pulmonary pathologies, specifically lung cancer metastasis is discussed in this dissertation, and remains an essential tool to study fundamental mechanisms in biology. Applications of the large crystal ribcage model can be closer to clinic and pharmaceutical industries to explore the effect of different ventilation strategies in disease, the effect of proning on lung ventilation and perfusion at alveolar scales, and improved pulmonary drug delivery, engagement, and uptake mechanics prior to full clinical studies
Towards a scalable and generalized end-to-end policy for autonomous driving
Get PDF2025The field of autonomous driving has witnessed a significant surge in the adoption of end-to-end frameworks, where raw sensor input is directly mapped to vehicle control signals. In contrast to the modular pipeline, which follows a sequential order of perception, prediction, and planning, the end-to-end approach offers advantages of integrated feature optimization for both perception and planning. However, it still faces several critical challenges, including scalability, interpretability, and robustness. Achieving scalable and generalized end-to-end policies for autonomous driving remains a fundamental challenge due to limitations in data diversity, distributional bias, and generalization to novel environments. Prior approaches, such as X-World (Zhang et al., 2021a) and ASSISTER (Huang et al., 2022), demonstrate suboptimal performance when deployed in diverse real-world settings, highlighting the need for more robust and scalable learning frameworks. In this work, we introduce a series of improvements across representation learning, policy learning, and structured sensorimotor training to enhance generalization and scalability in end-to-end driving policies.
First, we focus on learning scalable representations via pre-training. We propose NeMo (Huang et al., 2024), a neural volumetric world modeling approach that can be pre-trained in a self-supervised manner for image reconstruction and occupancy prediction tasks, benefiting scalable training and deployment paradigms such as imitation learning. We also introduce XVO (Lai et al., 2023), a semi-supervised learning method, pre-trained for multi-modal auxiliary tasks, i.e., segmentation, flow, depth, and audio prediction, to facilitate generalized representations for monocular Visual Odometry (VO) tasks across diverse datasets and settings. These techniques help improve spatial-temporal scene understanding across diverse conditions.
Second, we enhance policy learning through diverse supervision. To increase the supervision with new perspectives and maneuvers, we introduce LbW (Zhang and Ohn-Bar, 2021) which enables learning a driving policy from the demonstrations of other non-ego vehicles in a given scene without requiring full knowledge of neither the state nor expert actions. We further propose SelfD (Zhang et al., 2022b), an iterative semi-supervised training framework for learning scalable driving by utilizing large-scale unlabeled online monocular data. Both methods leverage diverse supervision for relieving the long-tailed maneuvers and domain shift issue and advance the underlying development process of safe and scalable autonomous vehicles.
Finally, we introduce structured learning strategies for sensorimotor agents. We present CaT (Zhang et al., 2023a), a novel distillation scheme that is trained with richer supervision in feature space and optimized via a student-paced coaching mechanism. We also introduce FeD (Zhang et al., 2024) that leverages advances in Large Language Models (LLMs) to provide corrective fine-grained feedback regarding the underlying reason behind driving prediction failures to improve robustness in complex driving scenarios.
Through extensive evaluation across multiple benchmarks, we demonstrate that our approaches significantly improves generalization, robustness, and scalability in end-to-end autonomous driving policies. Our findings highlight the potential of representation pre-training, diverse supervision, and structured learning to bridge the gap between simulation and real-world deployment, advancing the field toward truly scalable and adaptable autonomous driving systems