Alliance One Tobacco (Malawi)
Academic Research Repository at the Institute of Developing EconomiesNot a member yet
52305 research outputs found
Sort by
How to become a tardigrade? Combining paleontology and developmental biology to understand tardigrade body plan evolution
No full textTardigrades, also colloquially known as water bears or moss piglets, are a phylum of microscopic invertebrates that are characterized by having a segmented body plan and four pairs of lobe-like legs. Phylum Tardigrada is subdivided into the classes Heterotardigrada, which have high morphological disparity including various external cuticular specializations such as plates and spines, and Eutardigrada, which have a simpler plump-like appearance without external cuticular structures. Together with onychophorans and euarthropods, tardigrades form part of a larger clade known as Panarthropoda, with tardigrades having the distinction of being the only completely microscopic group in this context. Our current understanding on how tardigrades first evolved, and the processes leading to their morphological differences with the other panarthropods remains limited due to the scarcity on research that tackles these questions. In my dissertation, I performed an interdisciplinary approach that combined paleontological work and developmental genetics to shed light on the evolution of the tardigrade body plan. In Chapter 1, I described the youngest fossil tardigrade to date (in ~16 Ma amber) and showed that it had character combinations that are not present in extant relatives. This rejects the leading interpretation that tardigrades have remained morphologically static for millions of years. In Chapter 2, I redescribed the first two fossil tardigrades to be discovered, both preserved in a Cretaceous-age amber (~72 Mya) and resolved their phylogenetic relationships within tardigrades. This allowed me to effectively use one of them as a calibration point to calculate divergence time estimates (i.e., a proxy for when certain body plans existed) of the phylum and its major groups in deep time. In Chapter 3, I examined the gene expression patterns of distal limb patterning genes during embryonic leg development to infer which genes are involved in limb morphogenesis. Comparing the patterns to other panarthropod models suggest that tardigrade legs have a distal identity relative to the legs of onychophorans and euarthropods. In Chapter 4, I investigated the identity and fate of the eutardigrade embryonic tail – a feature that disappears during development. My results show that the tail gets internalized and potentially becomes the hindgut. Lastly, in Chapter 5, I redescribed the mid-Cambrian lobopodian Aysheaia pedunculata and provided support for its affinity as a stem-group tardigrade. This resolved relationship provided paleontological evidence for tardigrade miniaturization during their early evolutionary history, allowed the reconstruction of the ancestral tardigrade body plan, and revealed the morphological changes that have occurred in the lineage that led to the extant tardigrades. Overall, my dissertation utilized samples across the entire Phanerozoic to have different temporal perspectives – looking at the past and the present to ask how tardigrades have changed throughout their evolutionary history, and to uncover the macroevolutionary and genetic processes responsible for the stabilization of their distinctive body plan.Biology, Organismic and Evolutionar
Bioinspired Engineering: From Medical Devices to e-Nose Sensors
No full textNature rarely solves one problem at a time. It designs systems—delicate yet resilient, selective yet responsive—that operate under multiple constraints in complex environments. This dissertation draws from that ethos, presenting a body of work that bridges implantable medical devices and electronic chemical sensors, unified by a central theme: bioinspired engineering at the intersection of materials science, fluid dynamics, and computation.
In the first half of this work, I tackle long-standing challenges in biomedical implants—devices that must remain open yet resist fouling, deliver therapeutics yet prevent contamination, function reliably yet disrupt the body as little as possible. Using tympanostomy tubes as a case study, I develop subcapillary-scale conduits with curved geometries derived from capillary transport models, paired with liquid-infused surfaces that mimic the self-cleaning skins of pitcher plants. These tubes selectively transport desired fluids (e.g., therapeutics) while passively rejecting undesired ones (e.g., water, pathogens), with performance validated in vitro and in vivo. I extend these concepts to hydrocephalus catheters, where shear-driven self-cleaning and geometry-guided antifouling design prevent occlusion from cellular debris—a leading cause of implant failure. In both systems, I combine computational fluid modeling, materials engineering, and biological analogy to build devices that work with the body, not against it.
The second half of the thesis turns to the invisible world of chemical vapors, where biological olfaction offers a blueprint for high-dimensional, real-time sensing. Inspired by the sniffing behavior of mammals and the diverse receptor arrays in biological noses, I construct cross-reactive metal oxide sensor arrays integrated with machine learning models capable of classifying complex vapor mixtures. These e-nose platforms are deployed across domains—from food spoilage and air quality to breath-based diagnostics—achieving high specificity and adaptability by incorporating temporal signal processing, fluid-guided delivery, and adaptive sampling schemes. I postulate how feedback from flow simulations and physics-informed neural networks can guide sensor calibration and improve robustness, mimicking how natural systems sense dynamically and respond in real time.
While these domains may seem disparate—an ear tube and a chemical sensor—their design challenges and solutions are strikingly parallel. Both demand selectivity without complexity, resilience without rigidity, and the capacity to navigate the messy gradients of biology and the environment. By pairing natural design principles with modern fabrication, modeling, and data tools, this dissertation offers a cohesive framework for engineering systems that are not only inspired by nature, but perform with nature’s elegance and efficiency.
Ultimately, this work illustrates that the future of biomedical and environmental technologies will not come from any single field—but from the harmonious convergence of materials, computation, and mechanics, guided by the silent logic of biology. From implant to e-nose, this is a vision of devices that sense, respond, and endure—because they are designed like life itself.Engineering and Applied Sciences - Engineering Science
Intimately Bound: Injustice and the Foster System
No full textPolitical theorists have not paid sustained attention to the foster system or treated it as a political institution. Despite this, scholars in law and the social sciences, as well as a growing number of social movement advocates, have suggested that the U.S. and English foster systems are unjust. This dissertation characterises injustice in the widespread but relatively understudied institution of the foster system, proposes action-guiding ethical and policy principles to address these injustices, and suggests new normative ideals to govern the state’s intervention into intimate life.
To properly characterise injustice in the foster system, I argue that we must correct a conceptual oversight about the definitive power of the system and how it is typically employed. The foster system is an institution that engages in coercive relational intervention. More specifically, it enacts violence on our intimate relationships when it exercises its power to remove children from their caregivers. Much of the injustice in the system pertains to the misuse of this power, or, what I call, the capacity to employ relational violence.
Chapter one proposes the novel concept of relational violence which provides an ethical framework for evaluating moral harms to intimate relationships. Relational violence occurs when an external party damages an intimate bond by substantially disrupting an intimate relationship. Relational violence, I argue, is generally morally wrong and we have moral reasons to minimise its use where possible. In the foster system, however, poverty, housing insecurity, and racial prejudice can lead to unnecessary relational violence through child removals. The foster system thus emerges as an unjust institution that can compound the existing harms of racial and economic inequality with painful relational injustices.
Chapter two expands on how racism in the contemporary foster system damages intimate relationships. I argue that racial ideology specifically targets black parents in their capacities as parents. Racial ideology can suggest that the intimate bonds within black families are worth less than those within other families–or even that black intimate bonds are worth nothing at all. Critics have suggested that the foster system violates a “right to parent” or the family autonomy of black parents and families, and that it can violate a black child’s right to be raised in their racial group. Alternatively, some have even suggested that the system may under-intervene into black families. I show that my account of racial ideology as undervaluing, or totally devaluing, black intimate bonds complements some critiques of racism in the system while also addressing important counterarguments raised against others.
In chapter three I argue that despite the system’s purported goal of safeguarding children, the foster system’s demonstrable function–how it effectively operates in social and political life–is in fact a kind of poverty intervention. This poverty intervention can be direct or indirect. By this I mean that the foster system may intervene based on factors that impair parenting capacities but that may be either directly or indirectly caused by poverty. That the demonstrable function of the foster system is poverty intervention is itself concerning for theories of justice since poverty is a form of systemic injustice. But despite our moral obligations to correct for systemic injustice in society, state intervention within the foster system, I argue, should be determined primarily by our obligation to minimise damage to intimate bonds.
Finally, chapter four argues that our response to injustice in the foster system requires that the system aim to protect, where possible, the integrity of our intimate relationships. The concept of relational violence implies that we must aim to equally respect the moral value of other’s intimate bonds. This requirement is, I suggest, a necessary condition for the integrity of our intimate relationships. Nevertheless, respecting even this minimal account of intimate integrity requires substantial changes to law, policy, and practice in the foster system. It also demands that we clarify the extent and bounds of coercive relational intervention through the system, and that we work to make decision-making about intervention in the system more democratic. I conclude the chapter and the dissertation with some suggestions about what respect for intimate integrity would demand of the foster system.Governmen
New Tools, New Challenges: Navigating the Complexities of Digital Healthcare Delivery
No full textEmerging digital tools – such as telemedicine visits, remote monitoring, and patient portal messages – hold immense promise for improving healthcare access, quality, and efficiency. At the same time, they also introduce new challenges for healthcare organizations and policymakers. My dissertation sheds light on new dynamics brought about by three forms of digital care, providing practical guidance to managers on how to navigate them and policymakers on how to incentivize and enable effective use.
In Chapter 1, titled “From Rooms to Zooms: The Hidden Costs of Hybrid Work in Primary Care”, my co-authors and I examine hybrid primary care practices that offer both in-person and telemedicine care. My findings highlight new frictions that arise when virtual visits are incorporated into still predominately in-person clinic schedules. Intermixing the two modalities can lead to costly “modality switch” transitions that can negatively impact subsequent visits. Telemedicine visits following an in-person visit often see delayed starts; patients are 75% more likely to abandon the visit before being seen, and the visits that do occur are 25% less likely to begin on time. These disruptions also result in less comprehensive visits and a higher likelihood of after-hours work. Dedicated telemedicine-only blocks in provider schedules help avoid these costly transitions but can also lead to reduced capacity utilization when there is insufficient demand for telemedicine visits in that time window. Indeed, we find that telemedicine-only slots see a 10% lower booking rate relative to similar slots without such restrictions. Telemedicine visits are often framed as a useful tool for improving patient care access. However, we show that, depending on how they are incorporated into hybrid schedules, they can lead to negative care experiences, chaotic clinic days, and ironically even reductions in patient access. Our findings also demonstrate the tradeoffs of dedicated telemedicine blocks and highlight potential changes to managerial practices and clinical workflows to improve performance of hybrid practices.
In Chapter 2, titled “Practice-level Effects of Remote Physiologic Monitoring Adoption”, my co-authors and I leverage a 100% sample of Traditional Medicare claims data to study the practice level impacts of RPM adoption. Use of remote physiologic monitoring (RPM), the remote transmission of patient physiologic measures (e.g., blood pressure) to care teams, has grown rapidly in recent years. For practices, establishing an RPM program can increase revenue and improve patient care, but may also require substantial reorganization within the practice. No prior work has quantified the impact of RPM on practices. Using our Medicare claims dataset, we identified 754 primary care practices that began billing for RPM from 2019-2021. We find that, after these practices adopted RPM, Medicare revenue increased by 20.1% relative to similar matched non-adopting practices. This was driven by RPM billing as well as more outpatient visits and care management. While adopting practices had a 3.0% increase in their number of billing providers, the increase in revenue was predominantly driven by increased activity per provider. Adoption of RPM and resulting increases in visits for patients receiving RPM did not seem to come at the expense of other patients. Our results suggest that RPM holds promise as a tool for strengthening primary care and improving chronic disease management but also has the potential to substantially increase Medicare costs.
In Chapter 3, titled “The Doctor Won’t See You Now: Examining Drivers of Care Team Response to Patient Portal Messages”, my co-authors and I investigate the drivers of provider engagement with patient portal messages. Prior work has shown that when patients from historically disadvantaged groups (e.g., racial and ethnic minorities, those with lower socioeconomic status) send messages to their care teams, they are less likely to receive responses from physicians, seemingly driven by lower prioritization in the message triaging process. In this study we leverage natural language processing (NLP) tools to analyze the text of patient portal messages from a large academic health system. Our goal is to understand what drives these differences in care team response, enabling us to separate three potential mechanisms: differences in message content and the underlying request of the message (e.g., medication question, referral request), differences in the way the messages are written, and non-clinical bias. We find that, while message content is a significant predictor of care team response, it cannot explain observed differences across demographic groups. On the other hand, the way the message is written – including writing style characteristics such as length and formality – accounts for nearly half of the observed differences. Our findings identify a clear mechanism underlying disparities in care team response, highlighting avenues for mitigating them and deepening our understanding of care disparities more broadly.Health Polic
Cash Country: A Revolutionary Biography of the Tunisian Dinar
No full textCash Country is an ethnography of money in times of political upheavals. Following the 2011 popular uprisings in Tunisia and across the Arab region, I investigate how Tunisia’s national currency, the dinar, has become a public object. From widely commented-upon devaluations, media scandals on disappearing banknotes, currency trafficking at borders, new practices of hoarding money, and changing monetary policies, I argue that the Tunisian dinar structures revolutionary aspirations and their disenchantments, tying political horizons to unchanging economic conditions. Cash Country devises a methodology to “follow the money”, centering its materiality and circulations. I track the struggles that invest the dinar’s material forms, from cash whose visual aspect strives to mirror the transition between political regimes, to attempts by financial actors to digitalize currency. I follow the dinar from inside the Central Bank, showing how the institution transforms into the mediator of financial capitalism at home, to the nation’s borders where the dinar’s illicit circulations run in friction with transnational surveillance regimes.
Cash Country’s premise is the relation between money and its material forms, the exercise of materializing a universal media into a localized currency. I expand from the social theory of money which understands money as an object that evades definition because it exists mutually as a universal medium and a locally embedded form. I pay attention to social actors’ attempts to define money by bridging the gap between the idea of money – universal and commensurable – and its materialization into a national currency – depreciated and barely convertible. I argue that if the dinar has become a site of effervescence and interventions, it is because it articulates the scales people are caught in, between the frame of the national, where revolutionary transformations are imagined to take place, and the workings of global capitalism from where the imaginable gets produced. By following the social life of money, Cash Country writes a different story of uprisings and their afterlives in North Africa and the Middle East. Instead of assessing the successes or failures of revolutions, this dissertation highlights how struggles for freedom are struggles that invest the terms of capitalism. As political transitions give way to economic encroachment, Cash Country reveals how money operates as an object that structures horizons of possibility today.Middle Eastern Studies Committe
Ensemble Methods for Latent Structure Detection from Heterogeneous Genomic and Phenotypic Data
No full textDisentangling the hidden patterns within genomic and phenotypic data can improve our understanding of complex conditions. Recent methodological developments in statistics and machine learning have improved our ability to detect latent patterns in a variety of application areas; however, these methods are often unsuitable for some of the data types common to health and biomedical data. Likewise, many latent structure methods require prespecification of the dimensions of the latent space, which is typically unknown. In this work, we introduce three ensemble statistical and machine learning methods designed to fill in these gaps.
In Chapter 1, we introduce LACE-UP (LAtent Class analysis Ensembled with Umap and Pca), an ensemble machine learning method that outperforms gold-standard and oracle methods for clustering multidimensional binary data. When applied to dietary behavior data from the UK Biobank, LACE-UP uncovers interpretable dietary subtypes that are associated with lipid levels and cardiovascular risk. In Chapter 2, we introduce SEEK-VEC (Spectral Ensembling of topic models with Eigenscore for K-agnostic Vocabulary Embedding and Classification), a spectral ensemble topic modeling method for count data that yields prioritization scores and grouping scores that enable variable classification, pattern detection, and model diagnostics. We show through simulations that SEEK-VEC outperforms standard methods, particularly in weaker signal strength settings. We apply SEEK-VEC to single-cell gene expression data, food preference questionnaire data, and self-reported psychopathology symptom data, and show that the method uncovers meaningful insights across a broad range of contexts. In Chapter 3, we introduce SEEK-VFI (Spectral Ensembling of topic models with Eigenscore for K-agnostic Variable Feature Identification), an extension of SEEK-VEC that ranks genes with respect to their relevance to cell trajectory structure. We show that SEEK-VFI outperforms leading methods for differentiating between trajectory-relevant and uninformative genes, and we apply SEEK-VFI to several single-cell RNA expression datasets and demonstrate its ability to recover the true trajectory structure within the data.
This suite of methods, designed for non-continuous data, provide a lens into the latent structure underlying phenotypic and genomic data. These methods do not require the prespecification of the dimensions of the latent space and are robust to noise. Taken together, the promise of these methods and the development of similar methods in the future is a refined understanding of complex phenotypes and their underlying mechanisms, which in turn will improve diagnoses, prognoses, and care.Biostatistic
Differentiable Programming for Problems in Statistical Mechanics and Biophysics
No full textThis thesis explores how differentiable programming -- a paradigm that leverages automatic differentiation (AD) for scientific computing -- can be used to advance modeling and design in soft matter and biophysics. Traditional applied mathematics relies on hand-crafted models, approximations, and domain-specific numerics. However, recent advances in hardware acceleration and AD frameworks originally developed for deep learning have transformed the landscape of scientific computing, enabling exact and efficient computation of gradients in complex models.
I demonstrate how this computational shift enables novel capabilities across several domains. I first show how AD enables otherwise intractable analytical calculations. Specifically, I use AD to efficiently evaluate partition functions and assembly yields in systems of anisotropically interacting particles. I then apply this framework to compare calculations under existing models of protein-protein interactions with experimentally-determined assembly yields of de novo proteins. Inspired by this example of poor model generalization, I then focus on AD as an optimization tool for physics-based models. I devise a framework for directly differentiating the aforementioned assembly yield calculation, allowing me to fit protein force fields to target assembly yields via gradient-based optimization. I then extend these techniques to thermodynamic models of nucleic acids, showing that parameters in the popular "nearest neighbor" model describing secondary structure thermodynamics can be fit to data via gradient descent, enhancing predictive power. I also apply differentiable molecular dynamics to design functional colloidal systems.
For some physics-based calculations, direct differentiation for modeling or design is infeasible. One such cause is that a calculation is differentiable in principle but computationally prohibitive to unroll. In the face of this, I leverage and extend novel methods for stochastic gradient estimation to develop a framework for fitting coarse-grained force fields to experimental data. In the second limiting case, a calculation may be inherently discontinuous due to discrete control variables. One example of this is designing RNA sequences with respect to the aforementioned nearest neighbor model, for which I introduce an algorithm to compute the expected partition function over a probability distribution of RNA sequences, enabling gradient-based RNA design. Building on these advanced methods for stochastic gradient estimation and this probabilistic sequence representation, I develop a general method for inverse design in molecular simulations by introducing a notion of expected Hamiltonians. I demonstrate how this enables the rational design of intrinsically disordered proteins, DNA sequences, and even improved particle linking algorithms.
I conclude with a forward-looking perspective on promising applications of these methods, opportunities for future methods development, and proposed directions for novel interfaces between computation, mathematics, and physics.Engineering and Applied Sciences - Applied Mat
