36214 research outputs found
Sort by
Survey-Based Methodologies for Enhanced Assessment of Cause of Death
No full textThesis (Ph.D.)--University of Washington, 2025This dissertation explores several statistical challenges in cause-of-death (COD) assessment from verbal autopsy (VA) surveys—structured interviews with caregivers of the deceased in regions where traditional medical certification is unavailable. Despite their crucial role in mortality surveillance, VA data analysis is complicated by inconsistent age categorization, respondent burden from lengthy questionnaires, and potential biases in automated classification systems. The first project develops a Bayesian framework for reconciling inconsistent age categories across multiple VA data sources. We formulate age-disaggregated death counts as fully classified multinomial data and show that incorporating partially classified aggregated data can produce an improved Bayes estimator under the Kullback-Leibler (KL) loss. Under specific theoretical conditions, this approach calibrates data with different age structures to generate unified estimates of standardized age distributions. Through numerical studies and applications to real-world mortality data, we demonstrate the method's effectiveness in imputing incomplete classifications and guiding appropriate levels of age disaggregation. The second project adopts Bayesian active questionnaire design to optimize VA data collection processes. Using posterior-weighted KL information criteria and uncertainty-aware stopping rules, this approach sequentially selects questions to maximize information while minimizing respondent burden. Validation with gold-standard VA data shows comparable classification accuracy using substantially fewer questions, with implications for improved data collection efficiency. The third project presents a statistical framework for valid inference using predicted causes from VA narratives. By extending prediction-powered inference (PPI) to multinomial classification, we enable unbiased parameter estimation when using natural language processing models for COD classification. Cross-site validation demonstrates effective correction for transportability errors and highlights the distinction between predictive accuracy and inferential validity. The last project proposes and validates a proof-of-concept Bayesian mixture model for estimating cause-specific mortality with incomplete age stratification. Using age-mixing proportions within a Bayesian framework, this approach shows that incorporating partially observed age data improves estimation compared to discarding incomplete records. Analysis of demographic survey data from multiple countries reveals that the proposed approach generally yields more accurate cause-specific mortality estimates, with performance advantages varying by the actual age distribution of deaths. Together, these methodological innovations address fundamental challenges in survey-based mortality surveillance, with applications extending beyond COD assessment to broader problems of inference with incomplete or predicted data
Reducing Energy Demands in Greenhouse Farming: The Potential of Passive Solar Greenhouses in the U.S.
No full textThesis (Master's)--University of Washington, 2025Controlled environment agriculture has significantly enhanced agricultural productivity around the world. However, greenhouse farming operations in Western countries remain highly energy-intensive, with heating accounting for the majority of energy use. In the Netherlands, heating accounts for around 74% of total energy input. Passive solar greenhouses (PSGs) offer an energy-efficient alternative by utilizing direct sunlight as the sole heat source. Their design incorporates thermal mass, enabling high thermal gain and substantially reducing energy consumption. While PSGs are widely implemented in Asia, their adoption at a commercial scale in the United States has been limited. This study investigates the feasibility of implementing PSGs across diverse U.S. climate zones by evaluating their thermal performance and energy-saving potential. Four locations representing different climate zones were selected for case studies. A passive solar greenhouse model was used for energy and lighting simulations, testing various combinations of wall assemblies, glazing materials, shading, ventilation, and night insulation strategies.
Results indicate that PSGs increase the duration of ideal internal temperatures in all four locations. They are particularly effective in cold and marine climates, where they significantly raise interior temperatures without supplemental heating. Their performance is less effective in hot climates. Given that heating accounts for the largest portion of energy use in conventional greenhouses, PSGs show strong potential for cold regions. As more than half of the U.S. falls within cold and marine zones, PSGs offer a promising solution for year-round vegetable production with minimal energy input
South Seattle in Focus: Baseline Data for Equitable Transportation and Public Space Planning
No full textThis project documents non-motorized transportation (NMT) and public life patterns in South Seattle—an area historically underserved and understudied in local transportation planning— using SDOT’s Public Life Assessment framework. Our baseline findings for several sites, including the future Hillman City Light Rail Station and surrounding area will help guide equitable planning, investment, and improvements to support community mobility and vibrancy
Modeling and Algorithms for Nonconvex TrajectoryGeneration Problems: From Constraint Reformulations and First-Order Proximal Methods to Structure-Exploiting Convex Solvers
No full textThesis (Ph.D.)--University of Washington, 2025Trajectory generation plays a central role in modern Guidance, Navigation, and Control (GNC) systems by converting high-level mission objectives into reference trajectories that comply with system dynamics and task constraints. Among various methods, optimization-based formulation is favored for its ability to integrate dynamics, constraints, and performance objectives within a unified framework. Although various optimization-based trajectory generation methods have been successfully deployed in practice, many suffer from fundamental limitations that undermine their reliability. In particular, existing approaches often lack rigorous theoretical guarantees for convergence and constraint satisfaction, especially when implemented in discretized form. Furthermore, real-time deployment is frequently hindered by high computational costs and a heavy reliance on manual parameter tuning. To address these challenges, this dissertation presents three core contributions that improve the theoretical reliability and computational efficiency for trajectory generation. The first part of the dissertation focuses on a structured form of nonconvexity arising from thrust lower bounds, a setting where a method known as Lossless Convexification (LCvx) has been widely adopted in applications due to its empirical effectiveness. LCvx addresses this challenge by relaxing the original nonconvex problem as a convex optimization problem whose solution, under certain conditions, is also optimal for the nonconvex formulation. However, existing LCvx theory provides guarantees only for continuous-time optimal control problems, where trajectories and controls are modeled as functions over a continuous time domain. In contrast, practical implementations rely on discrete-time formulations, where the optimization variables are finite-dimensional vectors defined over temporal grids. This gap raises concerns about the theoretical soundness of applying LCvx in practical applications. In this dissertation, we extend LCvx theory to the discrete-time setting and establish formal guarantees that support its use in realistic implementations. In particular, we show that the solution to the convex problem resulting from LCvx satisfies the original nonconvex constraints up to a number of violations bounded by a linear function of the state dimension~, where the exact form of the bound may vary across different problems. The second part addresses trajectory generation under general nonconvex constraints. We first introduce a unified modeling and algorithmic framework that integrates prox-linear methods with exact penalty formulations. Moreover, due to discretization, classical trajectory generation algorithms typically guarantee constraint satisfaction only at grid points, and violations may inevitably occur between grid points. To address this issue, we propose a novel approach that incorporates an integral reformulation of the constraints into the optimization procedure, thereby ensuring constraint satisfaction over the entire time horizon. Lastly, to reduce the manual effort commonly required for parameter tuning, we design a proportional-integral (PI)-inspired autotuning scheme within this framework, which introduces a vectorized exact penalty comprising both linear and quadratic terms. After each prox-linear subproblem is solved, the penalty weights are adaptively updated: the linear term accumulates constraint violations across iterations, analogous to the integral part of PI, while the quadratic term responds directly to the current violation, corresponding to the proportional part. Theoretical guarantees and convergence analysis are provided for all methods introduced in this part. Finally, we propose Newton-PIPG, an efficient method for solving quadratic programming (QP) problems arising in optimal control, subject to additional set constraints. Such problems can serve as subproblems from general nonconvex trajectory generation algorithms. Newton-PIPG integrates the Proportional-Integral Projected Gradient (PIPG) method with the Newton method, thereby achieving both global convergence and local quadratic convergence. The PIPG method, an operator-splitting algorithm, seeks a fixed point of the PIPG operator. Under mild assumptions, we demonstrate that this operator is locally smooth, which enables the application of the Newton method to solve the corresponding nonlinear fixed-point equation. Furthermore, we prove that the linear system associated with the Newton method is locally nonsingular under strict complementarity conditions. To enhance computational efficiency, we developed a specialized matrix factorization technique that exploits the typical sparsity structure of optimal control problems and makes use of block Cholesky decomposition. Numerical experiments demonstrate that Newton-PIPG achieves high accuracy and reduces computation time, particularly in settings where feasibility is easily guaranteed
The Association of Enteral Protein Intake with Outcomes in Trauma ICU Patients
No full textThesis (Master's)--University of Washington, 2025Background: Critically ill trauma patients have distinct nutritional requirements due to increased catabolism and metabolic stress. Existing nutrition guidelines provide limited and outdated recommendations on optimal protein intake for this population. Recent studies suggest a potential dose-dependent harm associated with high protein intake in critically ill patients. We conducted a secondary analysis of a randomized clinical trial to investigate the relationship between early protein intake and ventilator-free days (VFDs), and to assess whether blood urea nitrogen (BUN)—a known marker of poor outcomes in ICU patients—mediates this relationship. Methods: This analysis included 329 trauma patients from a single-center randomized trial conducted between 2016 and 2021 at a Level 1 trauma center. The primary exposure was mean protein intake (g/kg/day) over the first week of ICU admission. The primary clinical outcome was VFDs, defined as days alive and free from mechanical ventilation within the first 28 days. Competing risks regression was used to analyze VFDs to calculate the subdistribution hazard of extubation, accounting for death as a competing event. A causal mediation analysis evaluated whether BUN mediated the relationship between protein intake and VFDs. The association between protein intake and secondary outcomes including acute respiratory distress syndrome (ARDS), ventilator-associated pneumonia (VAP), and aspiration were analyzed using logistic regression. Sensitivity analyses were performed for a subgroup of patients receiving ≥8kcal/kg/day. Results: Median patient age was 46 years (IQR: 30–59), 78% were male, and median injury severity score (ISS) was 34 (IQR: 26–43). Median protein intake was 1.6 g/kg/day (IQR: 1.0–2.0). Median VFDs was 14 (IQR 0–20). Each 1 g/kg/day of protein intake in the first week of ICU stay was associated with a significantly lower hazard of extubation (SHR 0.66; 95% CI: 0.53 to 0.81; p<0.001). Secondary outcomes, including ARDS, VAP, and aspiration, showed no significant associations with protein intake. Causal mediation analysis indicated that each additional 1 g/kg/day protein intake resulted in 3.53 fewer VFDs (95% CI: –4.81 to –2.30; p<0.001), with approximately 26% (95% CI: 12.0% to 41.0%) mediated by elevated BUN levels. Conclusions: In critically ill trauma patients, higher enteral protein intake early in ICU admission was associated with a lower hazard of extubation and fewer VFDs, partly due to elevations in BUN. These findings challenge current recommendations advocating high protein supplementation in trauma patients and highlight the need for further trials to define optimal protein dosing strategies tailored specifically to this high-risk population
The Radar and Microphysical Properties in the “Dendritic Growth” Layer of Winter Storms: Findings from the IMPACTS Field Campaign
No full textThesis (Master's)--University of Washington, 2025Radar scans of winter storms frequently show enhancements in equivalent radar reflectivity factor (Ze) within the -18˚C to -12˚C cloud layer, often referred to as the “Dendritic Growth Layer” (DGL). However, the microphysical processes responsible for these radar signatures remain poorly understood due to limited in-cloud in situ validation. This study leverages coordinated airborne radar and in situ observations collected during the NASA Investigation of Microphysics and Precipitation in Atlantic Coast-Threatening Snowstorms (IMPACTS) field campaign over the Northeastern U.S. We analyzed 591 vertical profiles of Ku-band Ze gradient (dZku/dz), each averaged over 10 km (~1 minute) segments and grouped into five clusters using a k-means clustering algorithm. Two of the five clusters exhibited local maxima in the magnitude of dZku/dz ≥ 10 dBZe/km and corresponding increases in Ku-Ka dual-frequency ratio (DFR) ≥ 1.5 dB within the DGL. Coincident in situ observations revealed enhanced particle growth with temperature for these clusters. However, habit imagery and relative humidity measurements showed that dendrites were largely absent and that the RHw supersaturation criterion for dendritic growth was unmet across all clusters. Instead, side planes and polycrystalline plates dominated with RHw subsaturated in the DGL. These findings suggest that the growth of plate-like polycrystals, not dendrites, was primarily responsible for observed radar enhancements in these IMPACTS cases, offering new insight into the microphysical drivers of radar signatures in winter storms
Conceptions of Mental Health and Well-being in the Indigenous Wiwa Community of Colombia
No full textThesis (Master's)--University of Washington, 2025Background: Indigenous populations worldwide face disproportionately high rates of suicide and mental health challenges. In Colombia, these issues are intensified by poverty, armed conflict, territorial exploitation, and limited access to culturally appropriate care. To address this issue, the Indigenous communities and the Colombian government developed the Guidelines for the Care of the Spiritual and Mental Harmonies of Indigenous Peoples and Communities. However, the Indigenous Wiwa community was not included in this process, and their perspectives on mental health remain largely unexplored. Methods: This qualitative, exploratory study aimed to understand how members of the Wiwa community who live in Kemakumake conceptualize mental health. Data was collected using a combination of photovoice, semi-structured interviews, and participant observation, prioritizing voices of the participants. Thirteen participants (11 community members and 2 community health providers) were recruited through snowball sampling to explore culturally grounded perspectives on mental health and well-being. An inductive thematic analysis was carried out using Dedoose software. Results: One main theme emerged from the photovoice process: “Madre Tierra as a foundation for wellbeing and the importance of cultural preservation”. Participants emphasized nature, sacred sites, and traditional practices as central to health. Three additional themes emerged from interviews and observation: (1) Ruama Sunica: Health from Thought, Body, Nature, and Spirituality; (2) Healing Through Tradition: The Role of the Mamo (spiritual leader) and Ancestral Practices; (3) Sources of Imbalance: Environmental Harm, Western Influences, and Marital Issues. Findings highlight the Wiwa’s holistic and relational understanding of mental health, centered on cultural identity, spirituality, and harmony with the environment. Conclusion: This study affirms the Wiwa’s holistic conception of mental health as rooted in spirituality, collective life, and harmony with nature. It highlights challenges between intercultural policy and Indigenous autonomy, particularly in remote contexts. Climate change and Western influence impacts emerge as urgent. Participatory, culturally grounded methodologies proved essential for ethical knowledge co-creation and decolonial inquiry
Understanding Risk Perceptions Regarding Cardiovascular Disease Among Young Adults in Gilgit, Pakistan
No full textThesis (Master's)--University of Washington, 2025Background: Cardiovascular disease (CVD) remains the leading cause of death worldwide, disproportionately impacting low- and middle-income countries (LMICs) such as Pakistan, where mortality rates associated with CVD exceed the global average. Recent epidemiological shifts show increasing CVD incidence among younger populations. In the province of Gilgit-Baltistan (GB), where health infrastructure remains limited, CVD-related morbidity and mortality among young adults has been rising in recent years. Contributing factors among this demographic include widespread behavioral risks such as tobacco use, unhealthy dietary habits, and low levels of physical activity, along with a high prevalence of undetected and unmanaged hypertension. Objective:
The study explores risk perceptions regarding CVD among young adults in Gilgit who have a baseline knowledge of CVD and its risk factors. It examines whether this knowledge translates into action and identifies perceived benefits and barriers to adopting CVD-preventive behaviors.
Methods: The study employed a qualitative design using the Health Belief Model (HBM) as a theoretical framework. Semi-structured interviews were conducted in Urdu with 31 undergraduate students aged 19–25, enrolled at Karakoram International University (KIU), Gilgit. Participants were recruited via snowball sampling. Data was analyzed through thematic analysis, combining deductive and inductive approaches, using ATLAS.ti software. Results:
Participants recognized CVD as a serious condition and expressed concern over its increasing incidence among young adults. However, most perceived their own risk as low, except those with a family history of CVD or personal health issues. Perceptions of CVD as a distant threat led to a low prioritization of preventive behaviors. Even among participants who acknowledged a higher personal risk, adopting healthy behaviors remained challenging. Barriers included academic stress, digital distractions, and religious beliefs. Social and familial influences served as both key motivators for behavior change and barriers in contexts where participants had limited control, such as shared family meals. Individual willpower emerged as an important internal driver, shaped by these external dynamics. Conclusion:
Our findings highlight the need to develop culturally relevant, youth-centered interventions that promote CVD prevention by addressing both individual perceptions and broader social determinants of behavior
Towards Generalizable Open-World Robot Manipulation by Training with Off-Domain Data
No full textThesis (Ph.D.)--University of Washington, 2025Achieving generalization in open-world robotic manipulation is a critical step toward deploying autonomous agents in dynamic, unstructured environments. However, learning manipulation skills that generalize to unseen objects, layouts, and natural language instructions remains challenging, particularly due to the scarcity and narrow coverage of real-world robot data. This dissertation explores how off-domain supervision—ranging from synthetic data to large-scale vision-language pretraining—can be harnessed to build scalable, generalist robot systems. We present three systems that progressively tackle generalization across different levels of the perception-to-action pipeline. First, we introduce DeepIM, a pose refinement framework based on iterative render-and-compare, which enables accurate 6D pose estimation using only RGB input. DeepIM demonstrates robust generalization to unseen objects and views, providing a foundation for geometry-aware manipulation. Next, we propose STOW, a discrete-frame segmentation and tracking method trained entirely on synthetic data. STOW exhibits strong sim-to-real transfer in cluttered warehouse environments by learning object-centric, temporally consistent representations, enabling robust multi-object scene understanding. Finally, we develop HAMSTER, a hierarchical vision-language-action model that integrates pretrained vision-language models with robot control via an intermediate abstraction of spatial sketch trajectories. HAMSTER enables the interpretation of diverse natural language instructions and their execution across varied semantic, geometric, and visual contexts. Together, these systems chart a path from model-based to model-free design in robotics, demonstrating that careful use of intermediate representations, modularity, and off-domain learning can bridge the gap between narrow robot deployments and open-world capability. By leveraging world knowledge and pretraining from large-scale datasets, this work contributes toward the long-term vision of scalable, generalist manipulation systems that adapt flexibly to real-world complexity
Neuropeptidergic Modulation of Canonical Reward Circuitry Underlying Escalation of Cocaine Consumption
No full textThesis (Ph.D.)--University of Washington, 2025Substance use disorder (SUD) and its related harms are a major public health concern in the United States and across the globe. Pharmacological treatment options are limited or nonexistent for many types of SUDs. Harm reductionist approaches to SUD, including pharmacological and non-pharmacological, have proven more effective at reducing individual and society impact of SUD than traditional abstinence only based approaches. However, our incomplete understanding of the neurobiology underlying the progression and sustainment of SUD hinders the development of new, more effective treatment. This dissertation focuses specifically on one specific SUD-like phenotype characterized by an increase in drug consumption over time (termed escalation). Preventing escalation and/or reversing escalation decreases harm through decreasing the likelihood of infection, overdose, and potential monetary/social loss of the individual with a SUD. Understanding the neurobiological basis of escalation will inform future therapeutic strategies. The neurotransmitter dopamine has a long history of being implicated in addictive drugs and there is evidence suggesting its causal relationship to escalation. Here, I will investigate two systems – dynorphin / kappa opioid receptors and corticotropin releasing factor – in their ability to modulate drug consumption, specifically escalation of cocaine self-administration. Both of these systems are capable of modulating dopamine, modulating each other, and have been implicated in the negative affect and withdrawal symptoms of SUD that are theorized to escalate drug consumption. Through pharmacological and gene editing techniques, I demonstrate in Chapter 2 that the kappa opioid receptor system, specifically expressed in ventral tegmental area neurons, is necessary for escalation of cocaine consumption. Using the same techniques, I demonstrate in Chapter 3 that the corticotropin releasing factor system in the nucleus accumbens, specifically CRF-R1, is not necessary for the development nor sustainment of escalation of cocaine consumption. Additionally, in Chapter 4 I provide a survey of cocaine self-administration in female versus male rats and provide evidence that this behavioral paradigm is capable of producing the SUD-like phenotype of escalation equally in females and males. These data provide further evidence for the dynorphin/kappa system to contribute to the β-process of the opponent process theory of SUD, independent of corticotropin releasing factor modulation through CRF-R1. I specifically implicating the dynorphin/kappa system of ventral tegmental area neurons that project to the nucleus accumbens. In the discussion, I theorize about downstream mechanisms and future experiments to investigate the temporal dynamics of dynorphin in the nucleus accumbens and its potential to modulate dopamine release during cocaine self-administration to escalate drug consumption