22689 research outputs found
Sort by
Process Analytical Technology Guided Scale Up of a Quench Crystallization Process with Focus on Impurity Minimization
Get PDFThe scale up of small molecule manufacturing processes presents many challenges due to the need for precise control over unit operations responsible for key properties that impact drug bioavailability, efficacy, and safety. Crystallization is critical since it directly affects purity, particle size, and form of the final product. Filtration and drying further influence physical crystal properties, batch processing times, yield, purity, and quality. To address process knowledge needs for scale up, process engineers have implemented process analytical technology (PAT) for in-line data analysis. This essay establishes the use of PAT in the development of a small molecule API for HIV treatment. Experiment goals included gathering information on impurity formation, nucleation, crystal growth, yield loss, and solvent removal through nitrogen blowdown and heated drying. Over several experiments the entire process was tracked by PAT. Results for a 40-50g scale run in a 500mL reaction vessel showed plate shaped crystals that nucleate after approximately 15% dosing of the reaction mixture to the quench solution, easy filtrations with low yield losses (0.4 - 1.11%), efficient blowdown in 4-6 hours, and final methanol and water contents of 0.03 weight percent (wt%) and 0.01wt% respectively. Bulk density and cake compressibility data gained during these experiments indicate that the large-scale pilot plant will require three filter dryer drops for one batch.
Concerns noticeable in these experiments included substantial amounts of encrustation in the crystallization vessel, low internal cake temperatures during vacuum drying, foaming during filtration, and an impurity present throughout the process that grows during drying. Several experiments focused on understanding this impurity, including determining that methanol is the source of impurity growth. Spiking studies indicated that higher drying temperatures and lower concentrations both correlate positively with impurity formation. This poses a potential issue at the industrial plant, which has lower efficiency in blowdown leading to more optimal conditions for impurity growth. To better understand this risk, along with encrustation and mixing risks, the process will be scaled up to 200g in a 2L vessel. Findings from this 200g scale up will be used to further inform a scheduled pilot plant run at the 250kg scale
RANS STUDY OF FULLY DEVELOPED WIND TURBINE ARRAY BOUNDARY LAYERS
Get PDFLarger wind farms have the potential to capture more wind and produce more electricity. However, when wind turbines are deployed in large arrays, their efficiency is degraded due to the complex interactions between turbines, the wind farm, and the atmospheric boundary layer (ABL). Understanding these interactions is therefore critical to improving wind farm performance. A fully developed flow regime is established when the length of a wind farm exceeds the height of the ABL by over an order of magnitude. In the past this limit has been studied using Large Eddy Simulations (LES). The present study aims to model the asymptotic limiting behavior of large wind farms using Reynolds Averaged Navier Stokes (RANS) simulations. RANS is a less expensive method compared to LES; however, no studies concentrate on the accuracy of RANS in modeling the fully developed asymptotic flow in wind farms. In this work, we conduct RANS simulations with a periodic array of wind turbines modeled using the Actuator Disc Model (ADM) for various wind-turbine arrangements, turbine loading factors, and surface roughness values for which prior LES results are available for comparison. The results are used to obtain the effective roughness length scales experienced by the ABL. These length scales are compared to the LES results and with an existing analytical model. Also, the vertical profiles of the horizontally averaged quantities are compared between RANS and LES to evaluate the accuracy of the RANS model for calculating the fully developed wind-turbine array boundary layer (FD-WTABL) in such wind farms. Next, results from three two-equation RANS models are compared with each other, with LES, and with experiments. Based on the observed trends, although computationally inexpensive, the Wilcox − model and the standard − model predict the atmospheric dynamics well at the horizontally averaged mean level and above the wind turbine. However, both these models miss important features of individual turbine wakes, such as the wake region which is excessively smeared out by this RANS approach. In contrast, the SST − model predicts the wake region better but fails to predict the log-law boundary layer structure. Present results may imply that RANS has difficulties in accurately matching both the wake and boundary layer structure of the FD-WTABL. Since for the WTABL both types of flow structures coexist, LES may be preferable even if more costly
Quantifying the Proportion of Dementia Risk Attributable to Hypertension, Diabetes, and Smoking: The Atherosclerosis Risk in Communities Neurocognitive Study
No full textHypertension, diabetes, and smoking are key modifiable vascular risk factors for dementia. The commonly reported fraction of dementia risk attributable to vascular disease in the US (hypertension, 8.8%; diabetes, 7.0%; smoking, 6.0%) is likely underestimated due to methodology that systematically underestimates population attributable risk when risk factors cluster together or interact. More valid methodology is needed to estimate dementia risk attributable to vascular causes that accounts for comorbid vascular risk factors and their age-dependent effects. Furthermore, because associations of vascular risk factors with dementia are diminished in the oldest old (>80 years), whether vascular risk factors confer risk of dementia in this group remains uncertain. Here, we address these research gaps by estimating the individual (diabetes and smoking) and summary (from hypertension, diabetes and smoking) population attributable fractions for dementia risk across midlife and late-life in the Atherosclerosis Risk in Communities Neurocognitive Study, a prospective cohort study of over 15,000 individuals recruited from four US communities with 35 years of follow-up. Expanding on the prior work of this investigator estimating the proportion of dementia risk attributable to hypertension in this cohort, the Specific Aims of this study were: 1) to quantify the fraction of incident dementia over 33 years attributable to diabetes and to smoking in midlife and early late-life; 2) to evaluate the summary contribution of three major midlife and late-life vascular risk factors to 33-year incident dementia; and 3) to investigate associations of late-life vascular risk factors with 11-year incident dementia in robust (not pre-frail/frail) older adults as compared to non-robust (pre-frail/frail) older adults
ASSESSING RELIABILITY OF CAUSAL MODELS OF TRANSCRIPTION
No full textTo begin building computational simulations of human cells, we will need a list of genes, a gene regulatory network (GRN) showing how the genes control one another, and a set of dynamic models that recapitulates activity over time. Our list of parts is making steady progress owing to new technologies that sequence DNA and reveal which regions are unpacked for use. However, the GRN and the dynamic models remain a challenge even a quarter century after the completion of the Human Genome Project. Using publicly available data, this work empirically evaluates a broad array of modern GRN inference approaches on two of their main functions: inferring direct regulators of transcription and predicting outcomes of genetic perturbations. Direct regulators are inferred using statistical independence tests on transcriptome data and are checked by combining genetic perturbations with assays of protein-to-DNA binding. Perturbation predictions are generated from diverse machine learning methods, and their adequacy for causal inference is tested using genetic perturbations that are not present in any algorithm’s training data. None of the algorithms tested on either task yield reliable results, with false discovery proportions far exceeding expected rates and with trivial baselines typically achieving lower prediction error than bespoke models. A key takeaway is that transcriptome data lack an essential statistical property, causal sufficiency, without which reliable causal networks cannot be inferred
Ultra-processed food consumption: classification, proteomic biomarkers, and clinical consequences
Get PDFFood processing techniques such as drying, cooking, and fermenting have long been valued for preserving food and enhancing its nutritional quality. Historically seen as beneficial, food processing was associated with safety and nutrient fortification. However, in recent decades, growing concerns have emerged about the role of highly processed foods in contributing to chronic disease. In response, classification systems like Nova have been developed to differentiate levels of processing. Ultra-processed foods (UPFs), in particular, have influenced nutrition research and dietary guidelines, though debates continue regarding their definitions, uses, and health impacts. This dissertation explores food classification systems, the association between UPF intake and chronic disease risk, and biological markers of UPF consumption.
First, we assessed the reliability of four food classification systems—Nova, IARC, IFIC, and UNC—using cross-sectional dietary data from NHANES. While these systems differed in categorizing processed foods, all consistently linked higher UPF intake to increased BMI and systemic inflammation.
Second, using proteomic data from the ARIC study, we identified eight plasma protein biomarkers associated with UPF consumption, some of which were prospectively linked to increased risks of coronary heart disease, chronic kidney disease, and all-cause mortality. These biomarkers provide insights into the biological effects of UPFs.
Third, we examined the prospective relationship between UPFs and diabetes risk in a large U.S. cohort. Higher UPF intake was associated with an increased risk of diabetes, partially mediated by BMI. Sugar-sweetened beverages and processed meats showed the strongest associations, while baked goods and ice cream had weaker or non-significant relationships.
Fourth, we explored the cross-sectional link between UPFs and subclinical atherosclerosis using carotid MRI data. Higher UPF intake was associated with increased carotid plaque burden, including greater total wall volume, lipid core volume, and segmental wall thickness, suggesting a pathway through which UPFs contribute to cardiovascular disease.
These findings highlight the impact of food classification systems on dietary assessment, the role of plasma biomarkers, and the potential benefits of reducing UPF intake to prevent diabetes and cardiovascular disease. This work advances knowledge on UPFs’ systemic health effects and informs future dietary and biomarker-based strategies for chronic disease prevention
Skin reinnervation by regeneration and collateral sprouting after peripheral nerve injury in mice
No full textPeripheral nerve injury (PNI) often leads to both sensory and motor impairments.
Following injury, nerve regeneration, in which injured neurons regrow under the guidance of Schwann cells, facilitates nerve reconnection with their original target. Apart from regeneration, collateral sprouting—a distinct mechanism involving the sprouting of adjacent intact nerve branches into the denervated territory—also contributes to the skin
reinnervation process. However, the relative contribution and temporal progression of these two processes remain poorly defined, particular in models involving complete nerve transection. To address this gap in knowledge, we utilized the sciatic nerve transection (SNT) model to investigate the temporal reinnervation pattern in denervated
skin regions. Animal behavioral assays were conducted to assess the pain phenotypes and functional recovery throughout the reinnervation process. Immunohistochemical staining and subtype-specific neuronal labeling were used to identify neuroanatomical changes in the mouse hind paw and dorsal ganglions (DRG). Our findings revealed partial axonal recovery in denervated skin territories in the mouse hind paw after SNT injury. CGRP immunoreactive peptidergic fibers and NF-H immunoreactive myelinated fibers exhibited continuous reinnervation in the denervated skin areas. Functional and anatomical assays performed after second surgeries indicated that both regenerating and collateral sprouting nerves contribute to reinnervation and modulate functional
outcomes. In parallel, increased expression of immune cell in denervated skin suggests potential roles for the cutaneous immune response in mediating skin reinnervation following SNT.
Altogether, our findings indicate that regeneration and collateral sprouting both contribute to skin reinnervation. These two processes may differ in timing, spatial reach, and fiber subtype involvement, but together they shape the functional outcome of
reinnervation. To specifically evaluate the component of reinnervation driven by collateral sprouting, future studies will need to be refined, for example by adopting alternative surgical approach, focusing on blocking regenerating injured axons at earlier time points, to minimize potential confounding effects from axonal regeneration
Leveraging the Natural Statistics of Speech Production to Sustain Temporal Precision in Brain-Computer Interfaces for Silent Speech
No full textThis dissertation addresses a central challenge in developing clinical electrocorticographic (ECoG) speech brain-computer interfaces (BCIs) for individuals with severe speech impairments: the absence of reliable outward speech indicators for users unable to vocalize. Traditional BCIs rely on recorded audio from the user’s own speech to guide model training, yet in silent speech contexts—where patients cannot consistently vocalize or produce intelligible utterances—neither speech timing nor phonetics can be directly measured. By leveraging the temporal statistics inherent in natural speech production and anchoring internal speech attempts to established timing patterns, this work develops standardized approaches for localizing and aligning ECoG signals during real-time neural speech synthesis without requiring a direct audio target. The early chapters investigate the impact of temporal imprecision on neural voice activity detection (nVAD) and speech synthesis, demonstrating how even minor misalignments critically undermine BCI performance. This work also derives first principles identifying the key temporal factors that drive model convergence during iterative alignment discovery for nVAD. These insights are applied to the development of Neural Audiovisual Synchronization (NAVSync) and Temporally Guided Dynamic Time Warping (TG-DTW), two complementary frameworks that address temporal imprecision through both behavioral and algorithmic strategies. NAVSync employs audiovisual cues to establish functional baselines and intuitive behavioral guides, while TG-DTW provides robust realignment to correct residual inaccuracies in silent speech contexts. Together, these methods enable continuous, real-time audible outputs without relying on direct microphone recordings. The penultimate chapter shifts to the technical implementation of low-latency software and hardware pipelines for neural signal streaming and processing, ensuring precise synchronization across repeated sessions and seamless adaptation to clinical environments. Beyond clinical applications, NAVSync and TG-DTW offer a platform for studying internal speech modulation by anchoring speech attempts to established temporal dynamics, illuminating how neural intent unfolds during silent speech. By integrating behavioral insights, advanced alignment strategies and real-time technical solutions, this dissertation advances the field toward a future where individuals with severe dysarthria or those effectively locked in can regain not only functional communication but also the spontaneity and expressiveness of natural spoken language
Cultivating New Spanish Knowledge: Experience and Exchange in the Spanish Colonial Garden, 1570-1620
Get PDFThis dissertation examines the development of botanical knowledge associated with the movement of plants between gardens in Spain and New Spain during the sixteenth and early seventeenth centuries. Focusing on four sites, namely the San Francisco-San José complex in Mexico City, the Hospital de la Santa Cruz in Oaxtepec, the Alcázar and nearby gardens in Seville, and the royal estates in Madrid, it relates changes in garden contents to shifts in local botanical thought that came with trans-Atlantic transfer. It also considers the changing role of the garden space in expressing and shaping beliefs about plants’ properties, including their medical, symbolic, and aesthetic value.
To examine these shifts, the dissertation examines printed texts on gardens, plants, and plant-related medicine produced at the sites in question, including Francisco Hernández’s Historia natural de la Nueva España, Gregorio López’s Tesoro de medicinas, and Gregorio de los Ríos’s Agricultura de jardines, among others. The plants described in these works are then traced through manuscript evidence from the sites and related institutions, including inventories, correspondence, purchase records, and drawn maps. Reference is also made to both European and Indigenous artistic and archaeological evidence, describing changes in garden contents over time alongside changes to the architecture and symbolism of gardens as a whole.
The final decades of the sixteenth century marked a turning point in the Spanish relationship to indigenous American knowledge about plants, shifting from a largely compilatory initial phase into a more synthetic one. As Spanish immigrants and their descendants interacted with American plants in gardens, they began to describe their nature and properties in ways that aimed to make indigenous American knowledge intelligible to a European-educated audience while also neutralizing elements that conflicted with a Christian worldview. In the other direction, gardens – particularly institutional gardens – were a key vector for the introduction of European plants and plant-related beliefs in what is now Mexico. These were adapted within a Nahua system of botanical knowledge that, despite official efforts to the contrary, would retain many pre-Christian elements into the seventeenth century and beyond
Label-Free Optical Biosensing for Malaria and Osteoarthritis
No full textThis thesis presents two complementary, label-free optical biosensing platforms—spatially offset Raman spectroscopy (SORS) and quantitative phase imaging (QPI)—and demonstrates their utility across disparate biomedical challenges. In Chapter 2, we establish a material-agnostic theoretical foundation for SORS in turbid media by deriving closed-form expressions for photon sampling depths and optimal source–detector geometries via Monte Carlo simulations. Building on this, Chapter 3 introduces pulse-correlated SORS, in which Raman acquisition is dynamically gated to the cardiac cycle. In silico results reveal that simple Fourier-based pulse correlation selectively amplifies blood-specific Raman features and markedly improves classification accuracy among uninfected, ring-stage, and gametocyte-stage malaria infections, reducing critical false negatives in advanced stages.
Chapter 4 translates the SORS framework into osteoarthritis research by mapping depth-dependent enzymatic degradation of bovine cartilage. By coupling Fick’s law diffusion models with the Monte Carlo-derived photon distribution, we obtain high-fidelity, zone-specific measurements of glycosaminoglycan depletion that correlate strongly with depth-resolved effective moduli from compression and shear testing. Finally, Chapter 5 introduces QPI as a label-free optical modality for malaria vector surveillance. We show that Plasmodium oocysts in mosquito midguts exhibit distinct refractive-index signatures that can be imaged without staining, and validate these findings through one-to-one registration with mercurochrome-stained brightfield microscopy.
Together, these studies demonstrate the power of label-free biophotonics to probe molecular and structural signatures noninvasively, whether through subsurface vibrational contrast or refractive-index mapping, and lay the groundwork for portable, high-throughput diagnostics in global health and musculoskeletal care. Future work will focus on experimental validation, melanin-compensation strategies, machine-learning–driven automation, and hybrid optical modalities to translate these approaches into point-of-care devices
PRINCIPAL-AGENT RELATIONSHIPS AND THE FATE OF CHINESE OVERSEAS INFRASTUCTURE PROJECTS:THREE CASES FROM MYANMAR
Get PDFThe dissertation seeks to address the longstanding question of what factors drives the fate of mega Chinese overseas infrastructure projects of Chinese state-owned enterprises (SOE). Using the theoretical framework of Principal-Agent relations, the dissertation examines the factors embedded within the state-SOE relations that affect the projects’ ability to sustain through local turbulences. The three cases of mega Chinese infrastructure projects in Myanmar during the country’s political reform and democratization between 2011 and 2015 present an ideal context for comparative studies given the controlled environment, the exhaustiveness of the cases, the shared experience of being suspended due to local concerns and the significant variations in the eventual outcomes of the projects: the Myitsone dam project has remained suspended until today; the Sino-Myanmar oil and gas pipelines successfully completed and have been in operation; and the Letpadaung copper mine was allowed to continue upon revisions to its project plan. Based on evidence collected through fifteen years of working on China-Myanmar relations and in-depth process-tracing, the finding is that the state plays a critical role in the management and negotiation of projects that run into trouble in a foreign country, and the relationship between the state and the SOEs—in which the projects are created in the first place—largely determines the resources the state is willing to allocate toward the rescue effort. Within that framework, the unity of the state, the compatibility of the goals between the state and the SOEs, the nature of the elite politics and relationship between the SOE leadership and the state leaders, and the budgetary resources are four key independent variables that directly affect the outcome. This research presents a useful comparison and complement to the existing literature that focuses on the exogeneity and externality of Chinese overseas economic projects. Instead of an “outside-in” analysis of recipient countries’ agency to influence the result of such projects, this dissertation takes the “inside-out” approach to examine how the internal dynamics of the projects affects the outcome of the projects