22689 research outputs found
Sort by
CALIFORNIA SEA LIONS’ (ZALOPHUS CALIFORNIANUS) POTENTIAL AS COASTAL HEALTH INDICATORS AND UMBRELLA SPECIES TO IMPROVE COASTAL MANAGEMENT
No full textThe California coastline is 3,427 miles of ecological diversity that is home to 26.3 million people. While there are countless native species that can be found in the coastal areas of California, the California sea lion (Zalophus californianus) is a coastal marine mammal species that is found statewide. Sea lions share the same habitats as humans, including the need for fish and clean places to live and play. It is this similarity in requirements that makes the California sea lion a candidate for an umbrella species for determining if the coast is healthy and if the policies that are used for coastal health and protection are working. Identifying an umbrella species is a species that represents ecosystems or locations completely; California sea lions are an umbrella species due to their population distribution, natural history, and integration with human populations. Umbrella species differ from indicator species in their roles for the environment; both are ecologically important species. An indicator species may be limited to being an indicator of one factor on the environment, such as water quality. The role of the umbrella species is greater in that it encompasses some of the same responsibilities of the indicator species, but by protecting the species, you protect others that depend on it as predator, prey, and an indicator. The change in the role of the sea lions can be reflected in a change in how policy advocates for coastal health. Currently, most policies come from the national level, with some from the state level; it is carried out by state and local governments. By simplifying how policy advocates for coastal health, how it is monitored, and focusing on one species that can represent ecosystems along the coast, it will be more efficient and nimbler to make modifications on protections and policy. Using sea lions as an umbrella species in combination with centralizing the way coastal health is monitored and managed with make decision making more effective, particularly if the California Coastal Commission becomes the central agency that acts as an information and policy clearinghouse. Changing the California Coastal Commission would be necessary for there to be one organization dedicated to the health of the coastal area, the people that reside there, issues that affect it, and the other ecosystems that call the coastal area home. Even though the California Coastal Commission is currently engaged in some coastal preservation, this ecosystem focused management would shift focus to having the sea lions being used as early warning systems for coastal health, the management of the habitats being proactive instead of reactive and making management more accessible and usable by members of the community
A COMPARISON OF NUTRIENT CONCENTRATIONS IN WETLANDS ACROSS LONG ISLAND, NEW YORK TO POPULATION DEMOGRAPHICS AND GEOGRAPHICAL CHARACTERISTICS
No full textWhile nitrates and phosphates are essential nutrients required for photosynthesis in aquatic ecosystems, excess levels can cause a chain reaction of ecological damage through a process called eutrophication. Eutrophication occurs gradually in a natural setting, but is accelerated due to human waste, excess fertilizer use, and other anthropogenic sources. In eutrophied waters, excess nitrates and phosphates cause algal blooms on the water’s surface, blocking light from reaching plants and animals underneath and eventually leading to the loss of dissolved oxygen and the potential death of aquatic organisms relying on that dissolved oxygen. These algal blooms, along with nitrate pollution itself, can be detrimental to human health and the local economy. This study sought out to measure nitrate and phosphate concentrations in wetlands across Long Island, New York, and compare these values to population density, income levels and other economic demographics, geographical characteristics, and the presence or absence of sewer systems. In regression analyses with population demographics, it was found that there is no relationship between nutrient concentrations and population density, along with no relationship between nutrient concentrations and economic factors. With the use of two-sampled t-tests, it was found that significantly higher nitrate concentrations were found in areas not covered by sewer systems as opposed to areas that were with 99% confidence, along with the same for phosphate concentrations with 90% confidence. It was also found that significantly higher phosphate concentrations were found in Suffolk County, which has less sewer system coverage than its counterpart, Nassau County. Finally, significantly higher phosphate concentrations were found on the south shore of the island as opposed to the north shore, which can be attributed to Long Island’s glacial origin
TRANSLATING CONTEXT-SPECIFIC FINDINGS INTO MIDDLE-RANGE INSIGHTS: EXAMPLES FROM REHABILITATION, ROAD SAFETY, AND INTEGRATION OF HEALTH CARE SERVICES
No full textStatement of the problem
This thesis is motivated by the growing burden of injuries, chronic diseases, and years lost due to disability. It explores three different topics – rehabilitation, road safety, and health care integration – and translates its findings into a series of middle-range outputs to catalyze future research. Aim one asks – why is rehabilitation seldom prioritized? Aim two asks – how, why, and under what conditions can technical assistance advance evidence-informed road safety? Aim three maps integration measurement
and its applications.
Methods
All aims utilize qualitative methods, drawing from health and social science theories. For aim one, agenda setting theories guided abductive analysis of interview data and literature to develop a policy framework for the prioritization of rehabilitation. For aim two, I used a realist evaluation methodology and multiple case study design to construct a program theory for the Bloomberg Philanthropies Initiative for Global Road Safety. For aim three, I conducted a scoping review of peer-reviewed literature using the Rainbow Model for Integrated Care for data extraction and analysis.
Results
Results emphasize the determining influence of ‘context’ in how health policies are prioritized, implemented, and evaluated. For rehabilitation, problem definition, governance arrangements (how national and transnational actors organize for collective action), and structural factors (health systems and national legacies) shape prioritization. In road safety, technical assistance successfully worked through city actors to strengthen road safety capabilities, but city and national contexts were determining influences on the scale of change. For integration, 99 included studies showcased the context-dependent nature of integration measurement.
Conclusion
Explicit attention to ‘contextual’ factors – whether to craft advocacy strategies for prioritization, align technical assistance programs to local implementation contexts, or to measure a health systems intervention – is critical to responding to the growing years lost due to disability. Theory building research approaches can incorporate contextual insights while developing middle-range outputs to advance a broader research agenda on these topics
OBSERVATIONAL CAUSAL INFERENCE FOR NETWORK DATA SETTINGS
No full textObservational causal inference (OCI) has shown significant promise in recent years, both as a tool for improving existing machine learning techniques and as an avenue to aid decision makers in applied areas, such as health and climate science. OCI relies on a key notion, identification, which links the counterfactual of interest to the observed data via a set of assumptions. Historically, OCI has relied on unrealistic assumptions, such as the ’no latent confounders’ assumption. To address this, Huang and Valtorta (2006) and Shpitser and Pearl (2006) provided sound and complete algorithms for identification of causal effects in causal directed acyclic graphs with latent variables. Nevertheless, these algorithms can only handle relatively simple causal queries.
In this dissertation, I will detail my contributions which generalize identification theory in key directions. I will describe theory which enables identification of causal effects when i) data do not satisfy the ’independent and identically distributed’ assumption, as in vaccine or social network data, and ii) the intervention of interest is a function of other model variables, as in off-line, off-policy learning, iii) when these two complicated settings intersect. Additionally, I will highlight some novel ways to conceive of interventions in networks. I will conclude with a discussion of future directions
Characterizing the COVID-19 Pandemic Among Indigenous Persons in the United States: COVID-19 Burden, SARS-CoV-2 Vaccine Effectiveness, and Risk Factors for Severe Disease and Post COVID-19 Conditions
No full textBackground: As of January 31, 2023, over 102.3 million COVID-19 cases and 1.1 million deaths have been reported in the United States (US). Indigenous persons account for <3% of the US population but have accounted for a disproportionate burden of COVID-19 morbidity and mortality and are underrepresented in COVID-19 data. This speaks to a history of systemic neglect and erasure, which has resulted in poor health outcomes, including those for COVID-19. We partnered with Indigenous communities to provide local data and fill an unmet need.
Methods: This dissertation leveraged data from an active, healthcare-facility based respiratory illness surveillance system conducted in five Indigenous communities in the US. Chapter 4 is a quantitative analysis of COVID-19-associated hospitalization incidence and risk factors for severe disease. In Chapter 5, a test-negative design (TND) case-control study was used to estimate effectiveness of COVID-19 vaccines among persons aged ≥12 years. In Chapter 6, acute COVID-19 illness is characterized, occurrence of post-COVID-19 conditions (PCC) is quantified, and risk factors for PCC were assessed among a longitudinal cohort of SARS-CoV-2-positive persons. Data were collected between January 1, 2021 and December 31, 2022.
Results: Hospitalization incidence was elevated among Indigenous persons, although comparisons to other race/ethnic groups was limited by lack of Indigenous-specific data from national and state-level systems. Risk factors were consistent with previously published reports and included older age, comorbid conditions, and being unvaccinated. COVID-19 vaccines demonstrated robust protection against COVID-19-hospitalizations and outpatient medical visits. PCC were observed in >50% of adults and almost 25% of children at 3-months post-acute infection; female sex, >5 symptoms during acute illness, hypertension, and SARS-CoV-2 variant predominance at the time of acute illness were associated with higher risk.
Conclusions: Access to locally-relevant data is critical to a community’s ability to respond appropriately to public health emergencies, such as the COVID-19 pandemic. The dearth of Indigenous data in national and state-level estimates fosters persistent inequities and undermines these communities’ ability to make evidence-based decisions regarding its members’ health and safety. This dissertation provides local data for five Indigenous communities in the US and may be used to inform culturally-grounded intervention strategies
Identifying metrics of greenwashing in the Chinese Green Bond market using quantitative and qualitative company data. Its implication for energy finance.
No full textThe following research looks at greenwashing in the Chinese Green Bond market. It tries to examine the extent of it by using parent company qualitative and quantitative information to find a predictive measure. It also looks at the implication of historic and current Green Bond Standards. The primary objective is to identify misleading labels on bond issuances
TREATMENT OF CLINICALLY RELEVANT, LARGE CRANIOFACIAL BONE DEFECTS USING POINT-OF-CARE STRATEGY AND ENGINEERED BIOMATERIALS
No full textThe craniofacial region is of great relevance as it provides unique identity to every individual and accommodates several vital tissues and sensory organs. Any major injury to the face compromises normal physiology, self-esteem and can be life threatening. Critical-sized craniofacial bone defects do not heal spontaneously and require surgical intervention. The current gold standard treatment is autologous bone grafts; however, this approach has many limitations. Consequently, tissue-engineered bone grafts (i.e., scaffolds) are emerging as a compelling alternative treatment strategy.
Although scaffolds have been extensively studied and characterized in the context of calvarial and mandibular bone defects, the midfacial region remains relatively unexplored. Factors like complex bone anatomy, intimate association with sensory organs and the presence of sinuses present unique challenges from a surgical standpoint. There is a lack of a robust strategy for the treatment of critical-sized midfacial bone defect in the clinic. As a result, the objective of this thesis was to bridge the knowledge gap and advance the field in this regard.
Through this work, we have tried to provide an end-to-end solution for the management of midfacial bone defects in the clinic. To this end, we first designed and validated a cell-based point-of-care (POC) strategy for treatment of critical-sized midfacial bone defects in a porcine model. Briefly, we treated 2 cm full-thickness, segmental, periorbital bone defects in skeletally mature pigs using 3D-printed scaffolds in combination with intraoperatively sourced autologous cells. Second, we developed an advanced imaging workflow to characterize the spatial bone regeneration patterns at periorbital defect sites treated with scaffolds. We demonstrated that post-operative geometric mismatch at defect sites favor orthotopic bone regeneration. Finally, based on our findings from the pre-clinical studies, we engineered a novel bio-responsive material to concomitantly degrade at a rate commensurate with bone regeneration
Practice-Oriented Privacy in Cryptography
No full textWhile formal cryptographic schemes can provide strong privacy guarantees, heuristic schemes that prioritize efficiency over formal rigor are often deployed in practice, which can result in privacy loss. Academic schemes that do receive rigorous attention often lack concrete efficiency or are difficult to implement. This creates tension between practice and research, leading to deployed privacy-preserving systems that are not backed by strong cryptographic guarantees.
To address this tension between practice and research, we propose a practice-oriented privacy approach, which focuses on designing systems with formal privacy models that can effectively map to real-world use cases. This approach includes analyzing existing privacy-preserving systems to measure their privacy guarantees and how they are used. Furthermore, it explores solutions in the literature and analyzes gaps in their models to design augmented systems that apply more clearly to practice.
We focus on two settings of privacy-preserving payments and communications. First, we introduce BlockSci, a software platform that can be used to perform analyses on the privacy and usage of blockchains. Specifically, we assess the privacy of the Dash cryptocurrency and analyze the velocity of cryptocurrencies, finding that Dash’s PrivateSend may still be vulnerable to clustering attacks and that a significant fraction of transactions on Bitcoin are “self-churn” transactions.
Next, we build a technique for reducing bandwidth in mixing cryptocurrencies, which suffer from a practical limitation: the size of the transaction growing linearly with the size of the anonymity set. Our proposed technique efficiently samples cover traffic from a finite and public set of known values, while deriving a compact description of the resulting transaction set. We show how this technique can be integrated with various currencies and different cover sampling distributions.
Finally, we look at the problem of establishing secure communication channels without access to a trusted public key infrastructure. We construct a scheme that uses network latency and reverse turing tests to detect the presence of eavesdroppers, prove our construction secure, and implement it on top of an existing communication protocol.
This line of work bridges the gap between theoretical cryptographic research and real-world deployments to bring better privacy-preserving schemes to end users
Optical Systems and Signal Processing for Sensing and Computing
No full textEver increasing complexity of modern science requires measurements at unprecedented time and spatial scales, throughput, sensitivity and speed which pose many physical and technological constraints that require novel solutions. Optical sensing technologies along with recent advances in signal processing has enabled measurements beyond seemingly fundamental limits. More specifically, in this work, I will discuss methods for overcoming various physical, electrical and sampling limits for on-chip spectroscopy, terahertz (THz) field detection, physically unclonable function (PUF) security, photon Doppler velocimetry (PDV), and neuromorphic computing. On-chip spectroscopy promises chemical and biological sensing in a small form factor but the small device size poses physical constraints on the spectrometer resolution and range which is surpassed here through compressed sensing. Similarly, THz field detection has applications in THz spectroscopy but cannot be achieved due to the limited speed of electronics. A novel method is proposed to surpass the bandwidth limit through indirect measurements of the field shape and inversion of the forward model through deep neural networks (DNNs). Machine learning (ML) algorithms have been useful in efficiently modeling photonic devices in general. More specifically, here photonic PUF security under various ML attacks is investigated. Photonic PUFs derive security from their pseudo-random optical response which depends on microscopic fabrication variations that are difficult to model. ML attacks are shown to partially break the degree of security which is quantified through a novel information theoretic metric that is found to depend on the device nonlinearity. Optical nonlinear effects are effective for surpassing limits in other applications as well. For example, I show that four-wave-mixing (FWM) in highly nonlinear fiber (HNLF) can be used to build a time lens which can slow down beat frequencies from PDVs to surpass electrical bandwidth limitations for measuring high velocity dynamic range. Notably, the time lens PDV (TL-PDV) makes PDV a viable option for measuring velocities from inertial confinement fusion. Lastly, a novel bio-inspired training algorithm is used to train a photonic extreme learning machine which promises to surpass electrical limits on computation. In conclusion, I show that seemingly fundamental limits can be surpassed through computational methods and nonlinear effects for optical systems. Conversely, optical systems can be used to surpass limits on computation through photonic neuromorphic system design and novel algorithms for using such systems