22689 research outputs found
Sort by
Resin Lifetime and Aging Behavior in Protein A Chromatography
Get PDFProtein A (ProA) chromatography is crucial to the manufacturing process of monoclonal antibody 1 (mAb1) in the pharmaceutical industry, however, ProA resin is very expensive and has a limited lifetime. The continuous use of chromatography resin can lead to decreased binding capacity, ligand degradation, resin fouling, and consequently compromised performance, eventually requiring the aged resin to be replaced with new resin. Extending resin lifetime can lead to significant cost and time savings. The mechanism causing age-related resin performance deterioration in the mAb1 ProA chromatography step has been investigated in past studies, but the primary mechanism of aging was not clearly identified. To interrogate resin lifetime, MabSelect resin that had been aged to its current lifetime at manufacturing scale, was aged an additional 80 cycles under normal manufacturing conditions at lab-scale. Step yield was determined for each cycle and ProA pools were sampled and tested for impurity clearance throughout the aging process. Dynamic binding capacity was measured every ten cycles and breakthrough curves were generated to investigate the primary mechanism of aging. Step yield was maintained through 80 cycles and impurity clearance remained consistent throughout the aging process. Based on step yield, the current resin lifetime for the mAb1 ProA step can be extended by 70 cycles with a safety margin of additional 10 cycles based on the study. Breakthrough curve analysis determined ligand degradation to be the primary mechanism of resin aging. Future experiments can investigate the contribution of impurity fouling in resin aging
Across the Burning Sands
No full textThis video essay explores the JHU Collection of Middle East-inspired Sheet Music
American Empire and Liberal International Order: From the Spanish-American War to the United Nations
No full textThis dissertation argues that the ideological and institutional foundations of US power emerged from a global imperial experience, and not primarily from a ‘domestic’ liberal tradition. While most International Relations (IR) literature on US power begins with postwar US hegemony, this project investigates how US experiments in overseas empire shaped plans for a liberal international economic order after World War II. Drawing on primary evidence sourced from a range of archives, I conduct three case studies of the formation of the US state’s liberal international ordering project from the birth of US empire in the Caribbean and the Pacific to the establishment of the United Nations. Each case study is anchored on a key actor – Charles Conant, Paul Reinsch, and Isaiah Bowman – whose intellectual influence brought them deep into the corridors of power. The case studies traverse US policy in Latin America, the Philippines, China, and beyond. Traveling between metropole and periphery, I demonstrate how these three practitioner-theorists brought their solutions to imperial problems home, turning them into foundational US policies. The project contributes to the IR literature on the roots of US hegemony, and re-centers the question of American empire in debates on the nature and sources of liberal international order
Race, Justice, and the Death Penalty
Get PDFThe word “justice” has implications on what it means for society to be fair. Yet, racial biases have infected justice since the founding of U.S. society and throughout its history, with race playing a major factor in how justice is carried out. These racial biases determine who is policed, disproportionately having a negative impact on Black, Indigenous, and people of color (BIPOC). This is further exacerbated by the death penalty, where these biases determine who lives and who dies. This argument emphasizes that the U.S. justice system makes the death penalty an unjust option for punishment in the current structure of the U.S. carceral system. Furthermore, by examining proposed reforms of the current system through the concept of bounded justice, this paper finds that these reforms to the current system lack the necessary features needed to make meaningful change. Meaningful change can only come when the U.S. carceral system dismantles the current justice system and sets up a system with foundations free of bias. Though this goal will likely take at least several decades to accomplish, the U.S. can start by suspending the death penalty
BUILDING RELATIONSHIPS AND STUDENT ENGAGEMENT: INCREASING SOCIAL STUDIES TEACHERS’ RACIAL LITERACY AND SELF-EFFICACY FOR CULTURALLY RESPONSIVE TEACHING PRACTICES THROUGH PEER COACHING AND INSTRUCTIONAL ROUNDS
Get PDFPrior research indicates a connection between culturally responsive teaching practices and student engagement in the classroom. Color-evasive pedagogy, which can negatively impact students’ understandings of content and course success, is also common in secondary social studies classrooms across the U.S. Factors contributing to color-evasive social studies pedagogy and the impact on student engagement were examined using Bronfenbrenner’s ecological systems theory. A mixed-methods needs assessment (N = 11) explored how high school social studies teachers understand culturally responsive pedagogy and what strategies they use to create safe classroom spaces, a strategy identified as culturally responsive and capable of increasing student engagement. Findings indicated that, although social studies teachers in this context understood the importance of incorporating culturally responsive teaching practices, there was a need for comprehensive professional development to identify specific strategies to incorporate culturally responsive practices specifically. A 5-week professional learning program utilizing peer coaching sessions and instructional round observations was designed to increase self- efficacy for culturally responsive teaching practices. The program was evaluated through the collection of qualitative data, including pre- and post-interviews, audio-recordings of peer coaching sessions, and a self-report survey. Findings indicate that the professional learning program had a high-level of participant engagement and was associated with increased self- efficacy for culturally responsive teaching practices in social studies classrooms. The purely qualitative data allowed for a thorough analysis and understanding of participants’ experiences. Limitations and implications for future research are discussed, acknowledging a need to connect teachers’ self-efficacy for culturally responsive teaching practices with increased student engagement and learning
ESSAYS ON HUMAN CAPITAL, LABOR MARKET POWER, AND EDUCATION POLICY
Get PDFIn the first chapter, I study the sources of employer monopsony power from the perspective of imperfect human capital transferability, defined as the portability of skills across occupations. I construct a measure of human capital transferability across occupations, and integrate this into a dynamic two-sided model of the labor market. Workers in this model make job-switch decisions over their life-cycle, cognizant of the depreciation in human capital value upon changing occupations. Imperfect transferability of skills gives firms some market power and allows them to reduce the wage, but it also makes it harder to replace skilled workers, reducing that power. Employers, in turn, post wage profiles that maximize their lifetime profits. I estimate the model using a matched employer-employee panel from Germany. I have three main findings. First, occupational switches are associated with significant wage penalties. Second, I find the life-cycle profile of wage markdown exhibits a U shape, where middle-aged workers suffer the smallest markdown. Third, I show that restoring perfect skill transferability lowers wage markdown for senior workers but increases that for younger workers. Further policy analyses show that a set of Active Labor Market Policies and education policies that feature both general education and vocational training have the potential to reduce labor market power.
The second chapter proposes a new approach to estimate the monopsony power of the labor market based on a forward-looking model of firm wage posting and worker job separation. Contrast to the literature, workers make job switch decisions based on firm-specific wage growth trajectories associated with different employers. The model is estimated using a matched employer employee panel data from Germany. The separation elasticity estimated from this model is greater than that from the conventional approach, suggesting that ignoring worker responses to heterogeneous wage growth rates lead to a potential overestimation of the actual monopsony power.
The third chapter studies the effects of the “Double First-class Construction” (DFC) initiative, a strategic program initiated by the Chinese government to enhance the competitiveness and quality of higher education in China, on the research output and coauthorship networks of selected universities. The difference-in-differences estimates find an increase in the number of economics publications associated with DFC universities post treatment compared to non-DFC universities. However, there is no increase in research quality. The dyadic treatment effect estimation finds a rise in coauthorship links involving DFC universities post treatment. These findings suggest a trend towards greater representation of publications from the selected DFC institutions
Quantifying success of multiple sclerosis lesion segmentation
No full textUnlike philosophy, art, or other methods of acquiring knowledge,
science is grounded by its use of empirical research. Scientists
conduct experiments intending to answer questions about their area of
research. At the cornerstone of all of this is evaluation. Scientific
evaluation is a structured reporting of the outcome of an experiment.
It needs to be a systematic application of scientific methods to
assess the implementation, improvement, and outcomes of an experiment.
In image processing, it is commonplace for the evaluation of
segmentation performance to be based on the statistic known as the
Sorensen-Dice Index (SDI). The previous use of the SDI amounts to a
rudimentary tool that limits the information it can expose.
In this work, we describe developments in the evaluation of medical
image segmentation, with application to multiple sclerosis (MS) white
matter lesion (WML) detection from magnetic resonance images. These
developments include refinements of the SDI to provide a finer grain
insight about the nature of the segmentation results being considered.
To facilitate this, we extended a classification scheme for object
correspondence in segmentations, to account for the situation where
more than one ground truth segmentation exists. Additionally, we have
generalized the definition of the SDI.
We demonstrate the additional knowledge that can be gleaned by these
improvements on multiple studies of MS WML detection with data from
multiple manual raters and across a wide range of algorithms
Visualizing the PHATE of Recurrent Neural Networks
Get PDFInterpreting the hidden representation of recurrent neural networks (RNNs) and how they evolve during training is crucial for advancing their applications and developing more effective models. We introduce Multiway Multiscale PHATE (MM-PHATE), a novel dimensionality reduction tool tailored for RNNs, designed specifically to visualize the dynamics of hidden units throughout their training phases. Unlike existing methodologies, MM-PHATE uniquely tracks and visualizes the progression of these dynamics as they relate to model performance across both intrinsic time steps and epochs. Our evaluations show that MM-PHATE captures meaningful dynamical structures, offering a deeper insight into the model's learning trajectory and overall performance compared to widely used methods such as PCA, t-SNE, and Isomap
DISPERSION, COLLISION AND COALESCENCE OF SPHERICAL BUBBLES IN TURBULENCE
No full textFrom natural phenomena such as the frothy wake behind a ship to various industrial processes like flotation, gas-liquid reactors, and wastewater treatment, bubble dispersion, collision and coalescence serve a crucial role by promoting bubble mixing and mass transfer, ultimately enhancing the efficiency of these systems. Despite their importance, limited experimental work on bubble dispersion, collision and coalescence in turbulent background flows can be found, due to challenges in optically capturing and tracking bubbles in turbulence in three dimensions (3D) at a high bubble density.
In this thesis, we first overcome these challenges by utilizing an intense turbulent generator consisting of a jet array and developing an advanced 3D Lagrangian high-density particle and bubble tracking algorithm. In order to generate the necessary turbulent conditions for our research, we conduct systematic experiments to understand how turbulence decay scales with the jet array configuration, including jet velocity, nozzle size, and nozzle spacings. A 3D Lagrangian particle tracking algorithm was developed based on the state-of-art Shake-the-Box method and is applied to the experimental data. This algorithm is extended to tracking bubbles at high density by incorporating the image cross-correlation method to resolve the challenge of precise bubble positioning in the presence of overlapping neighbor bubbles.
By employing these experimental techniques, we investigate the pair dispersion of both tracers and bubbles. Specifically, we studied turbulent pair dispersion, seeking to address the question of the universality of super-diffusive scaling, particularly Richardson's cubic scaling, which has been challenging to capture in previous experiments. Drawing upon our experimental observations, we developed a phenomenological model for the varying scalings in the super-diffusive stage, suggesting that the Richardson scaling can only be achieved at a small initial separation around three times the Kolmogorov length scale or at an infinite Reynolds number. This model also sheds light on the similar varying super-diffusive scaling observed in bubble pair dispersion. Additionally, we make an observation of a slowdown phenomenon during the transitional range of the bubble pair dispersion when two bubbles are initially in close proximity, which can be explained by the combination of vortex trapping and the hydrodynamic interaction between two bubbles.
By analyzing the trajectories of bubbles in close proximity, We can characterize two important quantities that were believed to be important for bubble coalescence, including the approach velocity and contact time, which determine the collision frequency and the coalescence efficiency, respectively. Both quantities show a distinct scaling from those predicted by classical hypotheses based on Kolmogorov theories. To account for these novel observations, we develop a generalized model for the ensemble average of the squared longitudinal relative velocity between two bubbles, namely the bubble structure function, as a function of the distance between two bubbles and the bubble diameter. %This model is based on the fact that bubble pairs obtain energy not only from turbulent eddies involving them but also from local eddies of similar bubble sizes.
By employing the bubble structure function, we derived the scaling of the bubble approach velocity, which aligns with the observations from our experiments.
Furthermore, the observation on contact time shows that the contact time is almost constant despite the variation in bubble size, suggesting that larger bubbles, which statistically approach each other at a faster velocity, experience a reduced chance of coalescing
Revealing the supernovae progenitors with early observations
No full textSupernova (SN) explosions are the violent end stages of stars. Even though these energetic events play critical roles in many areas of astronomy, from stellar physics and galaxy evolution to cosmology, there are still many open questions about their progenitors and physical explosion mechanisms. In my thesis, I will present my research on utilizing the earliest signals to constrain the progenitors and properties of several types of SNe. In particular, I utilize the high-cadence light curves from the Kepler and TESS telescopes that are ideal to monitor and characterize the fast-evolving light curve features within the first few days after supernovae explosions, the time when the signatures of the progenitors are not washed out yet by the expanding ejecta. First, I will present SN 2018agk and SN 2023bee, two SNe Ia observed with Kepler and TESS, respectively. SN 2018agk shows a smooth power-law rise, with no indication of any early excess. It is an SN Ia with an exquisite light curve that can be used as a prototype for future studies. In contrast, SN 2023bee shows a weak early excess in TESS, optical, and Swift UV light curves. We find that no progenitor and explosion model can explain the excess flux over the full wavelength range, and therefore both improvements in modelling and statistical study on a large sample of SNe Ia with early light curves are necessary to adequately disentangle the puzzle about the progenitor systems of SNe Ia. Second, I reveal the progenitor of SN 2021zby, an SN IIb with shock cooling peak covered in TESS, to be most likely a yellow supergiant with envelope mass of ~0.30-0.65 M⊙ and envelope radius of ~120-300 R⊙. Lastly, through the comparison with state-of-art spectroscopic evolution models, I constraint the progenitor of SN Ibn 2020nxt to be most likely a ≲4M⊙ He star that lost its ~1M⊙ He-rich envelope through binary interaction in the years preceding the explosion