22689 research outputs found
Sort by
Acceleration: Lessons from the Field
Get PDFThe Johns Hopkins Institute for Education Policy, supported by the Louis Calder Foundation and Chiefs for Change, has conducted the first in-depth case study of the use of acceleration strategies by public school districts. Through an analysis of documentation and assessment results, interviews with district leadership and school principals, and using multiple classroom observations, the Institute examined district policy and granular practices in math and ELA instruction at the middle school grade levels. Counter to expectations, we found that teachers in whole-class settings were closely following their respective high-quality curricula in both subjects – a key goal of acceleration. We found too that students were being regularly re-grouped for differential small-group and individualized online instruction – a key component of acceleration models. However, we found that the rigor of whole-class instruction was highly variable, even within the same school. We also found widespread skepticism on the part of principals as to the value of digital platform online learning. Finally, based on the assessment data and the observations we made, we can suggest the hypothesis that acceleration works best for students who are modestly behind in their learning (roughly up to a year), but that for those who are several (or more) grade levels behind, more drastic interventions will be necessary.Louis Calder Foundation and Chiefs for Chang
CONTROLLING EXCITONS IN VAN DER WAALS HETEROSTRUCTURES COMPOSED OF MOLECULAR SEMICONDUCTORS AND 2D TRANSITION METAL DICHALCOGENIDES
No full textControlling the energy and dynamics of excitons–coulombically-bound electron-hole pairs–is of major importance for many photonic, optoelectronic, and quantum devices based on semiconducting materials. Heterostructures composed of 2D atomic crystals offer exciting pathways to control excitons through misalignment between 2D layers, where the twist angle controls a so-called Moiré potential which can confine excitons. This dissertation introduces the use of a molecular crystal to tune the Moiré superlattice with symmetry, lattice period, and potential depth defined by the structure of a metal organic framework (MOF). To do this, we first synthesize high-quality 2D transition metal dichalcogenide (TMD) monolayers with chemical vapor deposition. We use the reaction between WO3 and NaCl to tune the metal flux and thus the morphology and disorder of monolayer WSe2 crystals. Then we study how excitons in a monolayer TMD are affected by the presence of a molecular layer on its sur- face. Using time-resolved and steady-state spectroscopy, we show that excitons in WSe2 transfer to the triplet state of rubrene, a molecular semiconductor which can undergo triplet fusion. Finally, we show that excitons in a TMD monolayer can be strongly altered by a 2D crystalline molecular layer. We demonstrate new assembly methods for MOF|TMD heterostructures through exfoliation and deposition of the MOF layer. We then use low-temperature spectroscopy to provide evidence of exciton confinement at the 2D MOF | 2D TMD interface due to the arrangement of the 2D MOF. This work paves the way for exciton control platforms which exploit the tunability of MOFs for passive and active control over TMD properties
MECHANOBIOLOGY OF CELLULAR SENESCENCE AND EV-MEDIATED CANCER METASTASIS
No full textTissue and cellular mechanics are inherently intertwined. To fully understand the complexity of one system without considering the intricacies of the other would be analogous to ignoring water in the process of erosion. In humans there are many age-associated pathologies that arise slowly with time - such as cancer and tissue fibrosis. The onset and progression of these conditions are characteristically linked to local physical forces. For example, how a solid cancer adapts to, and changes a local microenvironment represents a delicate balance between tissue remodeling and immune evasion. From a cellular perspective, mechanotransduction of environmental stimuli intricately governor physical and biochemical responses such as tumor cell proliferation and migration.
Here I take a biophysical approach to examining both cancer metastasis and cellular senescence. Using biomaterials designed to recapitulate pathophysiological conditions, I explore the role of physical features through the lens of three unique projects: 1) Characterization of cellular senescence at physiological scale, 2) Humanization of an anti-metastatic bispecific antibody therapy, 3) Mechano-regulation of cancer derived extracellular vesicles.
Applying mechanobiology techniques to these three projects, I first show that expression of intracellular proteins P16INK4a and P21CIP/WAF1 associated with cellular senescence are mechanosensitive. Substrate stiffness modulates protein expression further blurring the line between senescent and non-senescent cells. In my second project, we developed and characterized two new humanized bispecific constructs targeting the IL-6 and IL-8 signaling cascade. These constructs show a non-toxic anti-migratory response consistent with IL-6 and IL-8 inhibition. However, incorporation of an anti-IL-8R binding domain into our bispecific constructs induced significant immunogenic response. Finally, using tandem-mass-tag proteomics, we explore the role of matrix stiffness on cancer-derived extracellular vesicles. Vesicles isolated from cancer cells cultured on stiff tumor-like substrates show significantly different biodistribution behavior compared to soft vesicles in-vivo. Increased retention of vesicles derived from stiff substrates correlated with an increased abundance of integrins. These results suggest tumor cell vesicles from stiff tissue environments may be better suited to bind metastasis sites with potential downstream implications in premetastatic niche formation. Collectively, these projects demonstrate an intricate role for physical properties of the extracellular space; and their role on cellular mechanics that lead to tissue dysfunction
MICROSTRUCTURAL AND MECHANICAL CHARACTERIZATION OF STRUCTURAL METALS
Get PDFHerein the microstructural and mechanical evolution of two distinct materials was thoroughly characterized: Mg-1Zn-0.2Ca and ASTM A516 70 steel. The results generated from these studies are relevant to the advancement of processing/characterization of lightweight structural metals and to furthering the understanding of high temperature damage mechanisms relevant to the energy sector.
Mg-1Zn-0.2Ca was processed using rolling or conventional extrusion (CE), followed by Equal Channel Angular Extrusion (ECAE), utilizing different routes (which define the rotation of the billet in between extrusion passes), to identify the impact of processing on microstructure and mechanical properties. Preprocessing via rolling yielded a smaller, more homogeneous initial grain size than that produced by CE. After ECAE via routes 4Bc and 4A, the microstructural differences due to preprocessing were reduced and mechanical properties showed little variation between rolling and extrusion. The choice of ECAE route (4Bc or 4A) did not impact microstructural refinement. However, the 4Bc route led to enhanced ductility and higher ultimate tensile strengths, as compared to 4A.
Additionally, the impact of grain size on quasi-static and spall strengths of Mg-1Zn-0.2Ca was explored. Samples were processed to generate 3 target grain sizes: ~8, ~26, and ~57 μm. Grain size impacted quasi-static properties, as expected, and led to discontinuous yielding in finer-grained samples. Yet, spall strength did not vary significantly.
In the context of the energy sector, exposure to high temperatures/hydrogen pressures cause the microstructure/mechanical properties of pressure vessel steel to degrade due to high temperature hydrogen attack (HTHA), a significant concern for the petrochemical industry. Most guidelines regarding allowable environmental conditions are based on prior failures in-service. Thus, hydrogen creep data was generated on carbon steel (ASTM A516 70) at various temperatures (357-454°C), applied stresses (66-172 MPa), and hydrogen pressures (1.7-6.9 MPa H2). An unexpected creep plateau (after the tertiary regime) was observed in the low temperature samples. We provide evidence via creep analyses and interrupted tests suggesting that methane pressure, decarburization, hydrogen-defect interactions, and constrained growth of cavitated material all play significant roles in determining strain rates in the specific creep regimes. We also identify likely mechanisms that control creep rates in each regime
Frailty and Post-Transplant Delirium and Cognitive Impairment in Liver Recipients
No full textNeurocognitive complications are common among liver transplant candidates (i.e. hepatic encephalopathy) but also following liver transplant surgery. However, post-transplant neurocognitive complications remain poorly characterized, with highly variable estimates of the prevalence and limited characterization of post-transplant delirium and cognitive dysfunction within the first year post-transplant and long term. Indeed, it remains unclear if liver transplant recipients are at higher-than-expected risk of long-term cognitive dysfunction compared to the general population. It is also unknown whether there are modifiable risk factors for these outcomes, such as pre-transplant frailty. Physical frailty and cognitive frailty commonly coexist in liver transplant candidates, and frailty, delirium, and longer-term cognitive dysfunction have all been attributed to underlying inflammatory states. Furthermore, a liver disease-specific frailty index, the Liver Frailty Index, allows objectively identification of physical frailty among liver transplant patients, enabling its evaluation as a potentially modifiable risk factor for post-transplant neurocognitive outcomes.
To characterize post-liver transplant ICU delirium, we performed a retrospective cohort study of recipients enrolled in the Johns Hopkins site of the Functional Assessment in Liver Transplantation (FrAILT) study. Using granular clinical data, we found that delirium was highly prevalent, occurred early post-transplant, and was associated with pre-transplant ICU admission and higher Model for End-stage Liver Disease (MELD) score.
Next, we evaluated cognitive dysfunction 3-12 months post-transplant using Montreal Cognitive Assessment (MoCA) data collected at the Johns Hopkins Hospital and University of California San Francisco FrAILT sites. We characterized liver transplant recipients with and without post-transplant cognitive dysfunction. The strongest risk factor for post-transplant cognitive dysfunction was pre-transplant cognitive dysfunction.
Finally, we quantified long-term cognitive dysfunction among liver recipients ≥1 year post-transplant using a cognitive battery administered via telephone. We compared the characteristics of recipients with impaired, average, and superior cognitive performance. Cognitive domains in which recipients were most likely to have impaired performance included verbal learning, memory, and recognition and word and semantic retrieval.
Future plans include prospective evaluation of pre-transplant and post-transplant neurocognitive status in liver candidates and recipients. We hope to improve counseling of candidates regarding post-transplant neurocognitive complications, implementation of delirium prevention strategies, and recognition and supportive treatment of cognitive dysfunction
A PRE-AWARD REVIEW CHECKLIST: ENHANCING THE PRE-AWARD PROCESS AT CALTECH
Get PDFHigher education institutions that secure federal grants and agreements must implement internal controls to comply with each federal grant-making agency's unique demands, in accordance with their solicitations. Research administrators are integral in this process as they are tasked with the review of proposals, the upkeep of documentation, the fulfillment of administrative obligations, and proactive engagements with principal investigators and central offices to solve any pre-award complications. The aim of this project is to create a tool that helps to boosts the effectiveness and productivity of research administrators at Caltech throughout the proposal development process. The literature review sourced many institutional pre-award checklists, which captured agency-specific requirements. This information was integrated into the checklist to provide research administrators with a clear, structured checklist for conducting consistent pre-award review at Caltech
THE EFFECTS OF TERRESTRIAL AND EXTRATERRESTRIAL MECHANOSIGNALING ON CELLULAR AND TISSUE PROCESSES
Get PDFMechanobiological cues influence biological structures across all scales, from a single cell to a full
body. These cues are vital for normal biological processes, such as cell migration and division,
tissue structural composition, or astronaut orthostatic intolerance upon return to normal gravity.
These cues also influence disease pathogenesis and progression, such as in cancer. Here, I will
investigate mechanical influences on biological processes first at a single cell level, where we
investigate force dynamics as a glioblastoma cell migrates and invades its surroundings. This
investigation has impacts on high grade glioma invasion through the secondary structures of
Scherer in the brain. This is a cancer type that has stubbornly resisted many treatment efforts and
remains among the most lethal cancers with a 5-year survival of less than 10%. Next, we change
scales to the tissue scale, where we investigate the effects of spaceflight on human engineered
heart tissues. Spaceflight is well known to result in cardiovascular remodeling. This can have lethal
consequences: the only astronauts to have left low Earth orbit are 5 times as likely to die of
cardiovascular disease than the rest of the astronaut population. I show that spaceflight has
negative consequences on engineered cardiac tissues, which may result from oxidative stress and
mitochondrial dysfunction. Finally, both the spaceflight and cancer studies utilize polymeric
substrates which are known to absorb drug compounds, preventing accurate drug screenings. We
address this problem by showing that a polydimethylsiloxane-polyethylene glycol block
copolymer can prevent drug absorption in engineered tissue systems, enabling future studies to
conduct accurate drug screenings on the mechanobiological phenomena studied here. Together,
this work has impacts on cancer migration, spaceflight, and microphysiological systems in general
Evaluating Managed Care and Pricing Strategies in Medicare Advantage: Evidence from Macular Degeneration
No full textSpecific strategies used by Medicare Advantage insurers to generate healthcare efficiencies are incompletely understood. This body of work focuses on managed care and pricing strategies for drugs used to treat the clinical indication of wet macular degeneration. This is viewed an appropriate case study given that macular degeneration drugs are high-spending and have different value profiles, providing an opportunity to generate savings depending on the drug administered.
A 2019 policy permits Medicare Advantage insurers to implement step therapy on physician-administered drugs. Using the Medicare Advantage encounter data, and leveraging heterogeneity across Medicare Advantage insurers implementing step therapy and a pre-post design, Chapter 2 evaluates the causal impact of step therapy on insurer cost savings. The results report a 7.8% (95% CI: 4.9%-10.7% [p<0.001]) greater probability of being prescribed the lower-cost drug, bevacizumab, in the first administration due to step therapy. Chapter 3 evaluates the association of step therapy on longer-term physician practice patterns, including the extensive margin of overall treatment initiation and the intensive margins of timing of treatment initiation, treatment discontinuation, and medication switching within a treatment episode. This chapter finds no association between step therapy and longer-term treatment patterns.
Chapter 4 examines the variation in prices for aflibercept, which is one of the drugs used to treat macular degeneration. In traditional Medicare, the government reimburses physician-administered drugs at averages sales price, plus a fixed percentage, which is not subject to regulatory, market, or external forces and has been shown to misalign financial and clinical incentives. Private payers, including Medicare Advantage, are not required to use average sales price and can instead negotiate prices for physician administered drugs. Using the Hospital Transparency data, Chapter 4 finds that commercial, Medicare Advantage, and Medicaid managed care plans reimburse at higher prices compared to ASP for aflibercept. Within a insurer-hospital pair where one insurer negotiates with the same hospital for plans in different markets, commercial prices are 96% higher than Medicare Advantage prices and 84% higher than Medicaid managed care prices, on average. External forces may contribute to the ratios observed in the data
TOWARDS EFFICIENT METADATA-HIDING CRYPTOGRAPHY
Get PDFAlthough cryptography for confidential communications, computing on private data, and proving properties of secret values has seen much progress over recent years, there has been a noticeable lack of corresponding systems for hiding metadata. While many outside of the field believe that this is a problem that cannot be solved by technical means, to the credit of the cryptographic community, many cryptographic constructions have been proposed for various meta-data related problems, achieving strong security guarantees. However, these existing solutions either only work for limited settings or are too inefficient to implement in practice.
In this work, we propose new custom cryptographic primitives that can be used to hide three different types of metadata in three different settings: receiver identity in store-and-forward systems, sender identity in verifiable email communications, and user device location in offline finding networks. Moreover, we show that these primitives can be efficiently constructed and instantiated: at least one construction for each primitive has been implemented and micro-benchmarks are present, with computation and time complexity that appears reasonable for each given application. This work motivates the exploration of other types of efficient metadata hiding cryptography to solve practical, real-world problems
Statistical Learning via Stochastic Optimization under Data Privacy Considerations
Get PDFIn this dissertation, we study statistical learning, formulated as stochastic optimization problems, under modern constraints motivated by data privacy considerations. The goal is to understand the statistical and computational complexity in algorithm design for fundamental classes of problems.
The first part concerns differential privacy, which, in recent years, has emerged as the de-facto standard for privacy-preserving data analysis. We study design of differentially private algorithms for (a). supervised learning of linear predictors with convex losses, also known as convex generalized linear models, and (b). non-convex optimization, where the goal is to approximate stationary points of the risk function. Our derived guarantees for the proposed algorithms are, as of yet, the best-known, and in most cases, are shown to be nearly optimal, in the worst case.
The second part concerns the problem of machine unlearning. The goal here is to efficiently update a trained model under requests to unlearn a data point in the training dataset. We delve into the problem for widely-studied classes of convex losses: smooth/non-smooth settings and generalized linear models. We propose learning and corresponding unlearning algorithms, which are (non-trivially) accurate and efficient. Further, we extend our techniques, to unlearn general structured iterative procedures, and a streaming setting, where the unlearning requests arrive sequentially