Alliance One Tobacco (Malawi)
Academic Research Repository at the Institute of Developing EconomiesNot a member yet
52305 research outputs found
Sort by
Topographies of Grief
No full textTopographies of Grief is a collection of lyric essays that explore the relationship between grief, ethics, and the natural world. In conversation with scholars, writers, and artists such as Judith Butler, Virginia Woolf, and Nancy Holt, these writings are rooted in a conviction that grief attunes us to our fundamental interdependence.Author's Origina
Comparative Outcomes in Cervical Spine Surgery: A Dual Analysis of Revision Surgeries and Adjacent Segment Degeneration
No full textBackground
Cervical disc arthroplasty (CDA) is increasingly favored over anterior cervical discectomy and fusion (ACDF) for its motion-preserving capabilities and theoretical advantage in reducing adjacent segment degeneration (ASD). While evidence on revision surgery rates between CDA and ACDF remains inconsistent, the performance of individual CDA implants concerning ASD and related complications has received less comparative scrutiny. This study aimed to (1) compare the hazard and predictors of revision surgery between ACDF and CDA, and (2) evaluate the risk of ASD and complications across three commonly used CDA implants: Prodisc-C, Mobi-C, and M6-C.
Methods
We conducted a two-part retrospective cohort study at a single academic medical center. First, using propensity-score matching (2:1), 312 patients who underwent ACDF or CDA between January 2008 and June 2023 were analyzed to compare revision surgery rates and identify associated risk factors. Second, 70 CDA cases (17 M6-C, 21 Prodisc-C, 32 Mobi-C) performed by a single surgeon between 2008 and 2024 were reviewed for ASD development and implant-specific complications. Multivariable Cox regression was used to assess hazard ratios (HR) for revision surgery and ASD, adjusting for demographics, comorbidities, and operative characteristics.
Results
CDA was associated with a significantly lower hazard of revision surgery compared to ACDF (HR 0.36; 95% CI 0.017–0.77; p=0.01). Age, gender, diabetes, hypertension, heart disease, chronic steroid use, and opioid dependence were not significant predictors of revision risk. Two-level surgeries were not associated with increased revision hazard relative to one-level procedures (HR 0.78; 95% CI 0.48–1.53; p=0.48). Among CDA implants, the risk of developing ASD was not significantly different for M6-C (HR 1.94; 95% CI 0.19–20.14; p=0.58) or Mobi-C (HR 1.69; 95% CI 0.26–10.77; p=0.58) when compared to Prodisc-C. No demographic, lifestyle, or operative variables were associated with ASD risk, and no implant-specific complications showed statistically significant differences across implant types.
Conclusion
CDA is associated with a lower risk of revision surgery relative to ACDF, with no significant predictors identified among common clinical variables. Among CDA implants, Prodisc-C, Mobi-C, and M6-C demonstrated similar performance in terms of ASD risk and implant-related complications. These findings support the long-term durability of CDA over ACDF and suggest comparable outcomes across commonly used CDA implant designs.Medical Scienc
Optogenetic control and monitoring of ion transport in cells
No full textIon transport is a dynamic process that is important in signaling both between and within cells. Fluorescence-based measurements are well suited for investigations of ion transport due to their high sensitivity and high spatiotemporal resolution. Optogenetic actuators can locally perturb ion concentrations via combined optical and genetic targeting. Here we combine spectrally separate optogenetic control and monitoring to probe ion transport within cells. We apply these tools in neurons where ionic signals play many important roles across a wide range of space and time scales.
In Chapter 2 I investigate the effect on the intracellular pH of repeated neuronal stimulation with a blue-shifted optogenetic actuator, CheRiff, as measured with a red-shifted fluorescent pH sensor, pHoran4. We found that repeated CheRiff stimulation acidified neurons. The acidification was due to the channelrhodopsin CheRiff passing a proton current into the cell. To reduce this acidification, we characterized new channelrhodopsin variants with low proton permeability, and showed that they reduced the neuronal acidification and are viable alternatives for neuron stimulation.
In Chapter 3 I investigate diffusion of Ca2+ in dendrites, by combining a blue-shifted Ca2+ selective channelrhodopsin, CapChR2, with a red shifted fluorescent Ca2+ sensor, FR-GECO1c. Using patterned optical stimulation, we created small Ca2+ point sources in the dendrites of cultured neurons and imaged the diffusion of the Ca2+ ions along the dendrite. After the system reached steady-state characterized the length constants of the Ca2+ spread through the dendrites. We then did a paired measurement mapping the kinetic responses of the neuron cell body and dendrites to short wide-area Ca2+ perturbations. By combining this spatial and kinetic data, we estimated the effective diffusion coefficient for Ca2+ in dendrites. We then investigated the effects of Ca2+ efflux on Ca2+ spread and kinetics. We added a pharmacological blocker of NCX1 sodium-calcium exchanger and repeated the Ca2+ transport and homeostasis measurements. These experiments showed that the NCX1 exchanger had a larger effect on calcium handling in fine dendrite than near the cell body, an effect that could be explained by the difference in surface-to-volume ratio of these structures.
Finally, I discuss other optogenetic actuators and reporters that could be combined for control and monitoring of different signals of interest. I discuss the factors for choosing what tools to use together and how to ensure the proper functioning and interpretation. As more optogenetic actuators and reporters are created, there are ever more possibilities for probing important signaling pathways. However, it is also important to make sure that the molecular and optical tools are well understood and validated, as there can be artifacts which can affect cellular state and which may confound results, as seen in Chapter 2.Engineering and Applied Sciences - Applied Physic
Reflections on Solidarity: Greater Boston's Central America Solidarity Movement
No full textOver the course of the past year I embarked on a journey of experiential learning by spending over forty hours listening to the experiences and reflections of a group of activists. In particular, I interviewed and collected oral histories on the Central American Solidarity Movement from the perspective of seventeen different people in the Greater Boston Area: key organizers, politicians, and religious leaders that were active in the 1980s and 1990s.
According to Sharon Erickson Nepstad, the Central America Solidarity Movement “emerged in the early 1980s when US citizens sought ways to support Nicaraguans, Salvadorans, and Guatemalans who were fighting for human rights, economic justice, and democratic freedoms. Solidarity activists also worked to obstruct US political, economic, and military intervention in Central America.” I became interested in this social movement when I began mentoring a student at Somerville High School from Honduras. After hearing intimate details of how he emigrated to the United States as an unaccompanied minor, I began to research the “root causes” of migration. This led me to join an international human rights delegation to Honduras in August 2021. I have since been on delegations with the Witness for Peace Solidarity Collective (WFPSC), the Committee in Solidarity with the People of El Salvador (CISPES), and the Needham Guatemala Partnership to Cuba, El Salvador, and Guatemala, respectively.
From my involvement with these organizations, I developed a large network of older organizers and activists. As I took on a volunteer leadership role at Massachusetts Peace Action (MAPA), I realized that there is little to no published material on the Central America Solidarity Movement of the 1980s and 1990s in Greater Boston. This project served as an opportunity to research this important yet often overlooked social movement. It also provided me with the time to interview an older generation of activists, whose stories are being lost to aging memories and end-of-life.
The following two interviews are only a piece of the larger project. I chose to feature my two interviews with Dr. Jeanne Gallo (1938–2025) and Congressman Jim McGovern (1959–) because they were both early leaders of the Central America Solidarity Movement in the Greater Boston Area. As was characteristic of all of my interviews, I asked Dr. Gallo and Congressman McGovern three sets of questions. First, I asked my interviewees questions about their personal history, such basic facts as where and when they were born, where they grew up, who was in their family, etc. Second, I asked them questions about how they got involved in the movement and to think about the history, politics, and sociology of the movement overall. Finally, I asked my interviewees about what the movement meant to them, including some final reflections on the meaning of solidarity, the achievements and failures of the movement, and its lessons for the younger generations of activists today.
One key finding from my project is that the movement in Greater Boston contributed much more to the success of the overall trans-national movement than what has been previously acknowledged (e.g. Temporary Protected Status (TPS) was written by Jim McGovern and passed by Congressman Joe Moakley). In addition, spirituality for these activists, whether they identify as secular or religious, is rooted in a strong sense of social justice and human rights. Dr. Gallo put it best when she said that for her, “to be spiritual is to be a decent human being.” Their solidarity influences their spirituality, just as much as their spirituality informs their solidarity.
The following are verbatim transcripts and therefore are not altered to be grammatically correct. They aim to capture the flow and conversation style of the speaker. The audio files corresponding to the transcripts can be made available upon request.Author's Origina
TARGETING WOUND HEALING THROUGH PHARMACOLOGICAL & MODIFIABLE FACTORS: FROM PRECLINICAL EVIDENCE TO CLINICAL APPLICATION
No full textThis thesis investigates two complementary aspects of wound healing: the clinical impact of preoperative serum albumin levels on postoperative complications and the therapeutic potential of statins in enhancing wound repair based on preclinical evidence.
Paper 1 evaluates the association between hypoalbuminemia and postoperative outcomes in patients undergoing skin graft procedures. Using data from the ACS-NSQIP, a retrospective cohort study was conducted with propensity score matching. The analysis demonstrates that low preoperative serum albumin levels are significantly associated with increased odds of both wound-related and systemic postoperative complications. These findings highlight serum albumin as a potential biomarker, as well as a modifiable risk factor in surgical recovery.
Paper 2 presents a systematic review and meta-analysis of preclinical studies assessing the effects of statin therapy on wound healing. While statins are widely used for cardiovascular disease prevention, they also possess anti-inflammatory and pro-angiogenic properties that may promote tissue regeneration. By synthesizing findings from animal models, this study quantifies the effect of statins on wound closure and repair, suggesting a promising adjunctive role in wound management—one that warrants exploration in clinical settings, supporting a bench-to-bedside approach.
Together, these studies bridge clinical and preclinical perspectives on wound healing, emphasizing the importance of both nutritional status and pharmacological modulation in improving outcomes. They lay the groundwork for future research aimed at developing targeted interventions to enhance tissue repair and improve outcomes in surgical patients.Medical Scienc
Mechanisms governing inflammatory and invasive fibroblast pathology
No full textFibroblasts are mesenchymal cells abundantly present in the connective tissue stroma of most organs. Though classically regarded as passive bystander cells providing structural support to tissues, fibroblasts have increasingly emerged as active orchestrators of both tissue homeostasis and disease. From regulating immunity and organizing tissue microarchitecture to guiding the differentiation of neighboring cells, fibroblasts assume myriad roles in controlling normal tissue development, maintenance, and repair. By comparison, during diseases such as autoimmunity, cancer, and fibrosis, fibroblasts acquire aberrant inflammatory, fibrogenic, tissue degradative, or hyperplastic phenotypes that drive pathology. The persistence of pathologic fibroblast states in diseased tissues has been linked to treatment nonresponse and poor clinical outcomes, both in cancers and in chronic inflammatory diseases including rheumatoid arthritis (RA). Understanding the mechanisms governing pathologic fibroblast behavior will enable us to develop improved, fibroblast-targeted therapeutics.
The joint synovium, a connective tissue predominantly composed of fibroblasts, represents a valuable context in which to study fibroblast behavior and function. During RA, synovial fibroblasts dramatically expand in number and become pathologically activated to drive both chronic joint inflammation and the invasion and destruction of surrounding cartilage and bone. Critically, disease-associated fibroblasts are enriched in the synovia of the nearly 50% of RA patients who fail to achieve remission with conventional therapies. There is a sparse understanding of the mechanisms driving pathologic fibroblast activation and function. In this thesis, we applied a combination of experimental, computational, and functional genomics approaches to define new regulators of inflammatory and invasive fibroblast behavior.
1. First, we found that the Wnt signaling pathway, widely regarded as a developmental morphogen, functions as an unexpected and potent driver of inflammatory fibroblast activation in RA. Inhibition of Wnt signaling attenuates inflammatory disease in a murine model of arthritis and may represent a promising fibroblast-directed therapeutic strategy for RA.
2. Next, we examined how transcription factors regulate the inflammatory and invasive functions of fibroblasts. We observed that rather than exhibiting maximal inflammatory and invasive states simultaneously, synovial fibroblasts in RA are polarized to become either more inflammatory or more invasive. We discovered that the transcription factor ARID5B drives this divergence by blunting the inflammatory activation of fibroblasts while augmenting their invasiveness.
3. Finally, we performed a genetic deletion screen in primary synovial fibroblasts derived from RA patients to identify new regulators of pathologic fibroblast behavior. We characterized the roles of one integrin and two transcription factors in modulating key inflammatory and degradative factors produced by pathologically activated fibroblasts.
Together, our work presents a set of key insights into inflammatory and invasive fibroblast regulation that may inform the development of fibroblast-targeting therapies for RA and other chronic inflammatory diseases.Immunolog
The Context-Aware Quantization Design Space: Unlocking Scalable Training and Inference for Large AI Models
No full textThe rapid development of large AI models capable of remarkable performance in a diverse array of complex tasks has inspired its widespread application and deployment, making the demand for scalable AI models that support fast training and inference increasingly intense. Model quantization has emerged as a critical and widely applied technique for reducing the computational and memory demands of training and inference of large deep learning models. Recently, advances in GPU tensor cores for acceleration of low-precision floating point computation have driven adoption of quantized training, in addition to quantization at inference time. In tandem, the emergence of new architectures like state space models as alternatives to transformers and quadratic attention demand a diversification of our understanding of quantization dynamics beyond ad-hoc, model-specific solutions. For a given model, dataset, and downstream task--the context of quantization--we are faced with a combinatorially large and complex design space, within which any choice can have drastic implications on the stability of quantized training and accuracy degradation at inference time. This thesis proposes advances towards a unified framework for structuring the problem space of quantization, evaluating and realizing the computational gains of quantization, and for exhaustively and comprehensively characterizing quantization in a context-aware manner. We define two design principles--systems performance and model performance--as foundational objectives for quantized optimization and inference, and propose evaluative metrics and diagnostic experiments for understanding quantization dynamics from every dimension for a wide distribution of data regimes. As a proof of concept of our framework, we conduct an empirical study of FP8 quantized training of the Mamba-2 state space model, revealing new insights into the impact of quantization on its numerical stability, gradient norm dynamics, temporal and layer-wise quantization error propagation, and catastrophic degradation phenomena.Computer Scienc
Misuse of AI Through Adversarial Attacks
No full textThe advent of new generative AI models, such as OpenAI’s GPT-4, has seen a huge increase in AI capabilities. With this, there has been a corresponding increase in the ways in which such models can be used to commit harm by malicious actors, which we refer to as misuse. One such method to misuse models is through adversarial attacks. Informally, these are inputs to AI models that cause said models to act in ways that the developers did not intend. Most AI systems require users to, at the very least, have access to their input. Thus, if adversarial attacks for misuse exist, most systems are vulnerable. This makes adversarial attacks a natural and important setting to study AI misuse. The central question of this thesis is: can harm be committed with current AI models through adversarial attacks? We decisively answer this in the affirmative, showing that across a wide range of settings, malicious users can craft inputs that cause current AI models to misbehave in alarming and harmful ways.
In Chapter 2, we introduce the novel Interaction Context Framework, a collection of formal definitions for AI systems, model behaviors, and adversarial attacks, specifically designed to aid in the study of AI misuse. Notably, this is the first formalism to unify three previously disparate notions of adversarial attacks: adversarial attacks to discriminative models, prompt injection attacks to generative models, and jailbreaking of generative models.
In Chapter 3, we explore adversarial attacks when the user has access to the underlying AI model. We present the novel Image Behavior Matching Algorithm for creating images that force multimodal LLMs to display four different types of harmful behaviors. We also extend existing text-based attacks to these behaviors.
In Chapter 4, we consider the setting in which users can only access the AI’s input and output. We create Tensor Trust, an online game that pits players against one another in an adversarial attack and defense contest. Using this, we collect and release a dataset of 128,808 adversarial attacks and 46,457 defenses, as well as two corresponding robustness benchmarks. We evaluate current AI models on these benchmarks and find all perform poorly.
Work done in Chapter 3 was conducted jointly with Euan Ong, Professor Stuart Russell, and Scott Emmons. Work done in Chapter 4 was conducted jointly with Sam Toyer, Olivia Watkins, Ethan Mendes, Justin Svegliato, Tiffany Wang, Isaac Ong, Karim Elmaaroufi, Pieter Abbeel, Trevor Darrell, Alan Ritter, and Stuart Russell.Computer Scienc
A novel biomarker to identify atherosclerotic cardiovascular disease in people with HIV and assessment of clinical and ECG features associated with incident sudden cardiac or undetermined death: Insights from the Randomized Trial to Prevent Vascular Events in HIV (REPRIEVE)
No full textThe two manuscripts presented for this thesis have both analytical and thematic content
links that are consistent with: 1. addressing important scientific questions in a historically
underserved/under recognized high cardiovascular risk primary prevention population, and
2. Demonstrating the student’s facility with analysis and interpretation of clinical trial data
using dieerent study designs for these secondary analyses. These analyses of the
Randomized Trial to Prevent Vascular Events in HIV (REPRIEVE) also utilize the clinical trial
expertise of the student’s mentors, one who is the study principal investigator (SG) and the
other who was the study endpoint committee chairman (SW) who also serves in this
capacity for many phase III cardiovascular clinical trials managed through the Harvard
based academic research organization, the TIMI study group.
The overarching theme is that people with HIV (PWH) are at increased risk of
cardiovascular disease even in the present era of excellent viral suppression with antiviral
therapy.(1) The first manuscript utilizes a cross-sectional design to evaluate the
associations of “high-risk” coronary atherosclerosis imaged by CT coronary angiography
(CTCA) in a REPRIEVE sub study with the cardiac specific circulating protein, cardiac
troponin T (cTnT) measured by high-sensitive assay.(2, 3) We further sought to understand if
this biomarker could augment discrimination between those participants with and without
these plaque features.(3) Next utilizing a longitudinal study design we investigated clinical
and ECG features associated with carefully adjudicated sudden cardiac and undetermined
deaths highlighting the importance of meticulous adjudication to characterize the
dieerences in the clinical phenotypes in participants with these outcomes.Medical Scienc
