SACM - United States of America
Permanent URI for this collectionhttps://drepo.sdl.edu.sa/handle/20.500.14154/9668
Browse
12818 results
Search Results
Item Restricted GUIDELINE-DIRECTED LIPID-LOWERING THERAPY IN U.S. ADULTS: TRENDS IN UTILIZATION, TREATMENT PATTERNS, DISEASE BURDEN, AND HEALTHCARE EXPENDITURES(Saudi Digital Library, 2026) Alharbi, Abdulrahman; Sherbeny, FatimahBackground. Lipid-lowering therapy (LLT) is among the most widely prescribed pharmacologic interventions in the United States and plays a central role in the prevention and management of atherosclerotic cardiovascular disease. The 2018–2023 period spans the maturation of the 2018 ACC/AHA cholesterol guideline, continued generic market expansion, and the COVID-19 pandemic, yet contemporary nationally representative evidence on utilization, treatment intensity, disease burden, and expenditures among LLT users remains limited. Objective. To characterize trends in LLT utilization, statin intensity, comorbidity burden, payer-stratified expenditures, and predictors of high-intensity statin prescribing among U.S. adults aged 40 years and older who used LLT. Methods. A repeated cross-sectional analysis of the Medical Expenditure Panel Survey (MEPS) 2018–2023 was conducted, restricted to adults aged ≥40 years reporting at least one LLT fill (unweighted N = 25,662; weighted ≈ 281.5 million person-years observations). Survey-weighted estimates were generated in Stata 19 using svy procedures. Trends were tested with survey-weighted regression; predictors of high-intensity statin use were modeled with survey-weighted logistic regression, and expenditures with generalized linear models (gamma family, log link). Results. Atorvastatin use rose from 51.2% to 56.9% of LLT users and rosuvastatin nearly doubled from 11.5% to 22.8% (both p < .001), whereas simvastatin (23.2% to 12.7%), pravastatin (12.0% to 6.2%), and lovastatin (4.5% to 2.3%) declined (all p < .001). Ezetimibe use increased modestly from 3.4% to 5.7% but remained low. High-intensity statin therapy rose from 31.9% to 40.8% (p < .001). Total national LLT expenditures fell 71.4%, from $17.2 billion to $4.92 billion, with private-payer spending declining most steeply (−82.1%) and Medicare remaining the largest single payer. Higher-intensity prescribing was independently associated with cardiovascular events (adjusted odds ratio [aOR] = 2.1), higher Charlson comorbidity burden (CCI ≥3: aOR = 2.04), public insurance (aOR = 1.16), and later survey year (2023 vs. 2018: aOR = 1.54), and inversely associated with female sex (aOR = 0.62) and age ≥75 years (aOR = 0.76). Conclusions. Between 2018 and 2023, LLT prescribing shifted toward higher-potency agents and higher-intensity statin therapy while aggregate spending declined sharply, driven largely by generic substitution. Persistent underutilization of ezetimibe and demographic disparities in high-intensity prescribing identify clear targets for guideline-directed quality improvement.9 0Item Restricted A COMPARISON OF THE POWER OF NONPARAMETRIC TESTS FOR LOCATION AND SCALE IN THE TWO-SAMPLE PROBLE(Saudi Digital Library, 2026) Aldosry, Fahad; Rhonda, MagelThis dissertation compares six individual nonparametric two-sample tests (AnsariBradley, Siegel-Tukey, Savage, Hollander, Wald-Wolfowitz, and Mann-Whitney) and five combined location-scale statistics (AB+MW, ST+MW, H+MW, WW+MW, and Savage+MW), of which ST+MW and Savage+MW are proposed as new contributions. The Savage+MW statistic uses a divisor of 2 rather than the standard √2 because the Savage and Mann-Whitney rank statistics are positively correlated under H₀, the variance of their sum is at most 4, so this choice keeps the combined variance at or below 1 for any non-negative correlation. Monte Carlo simulation in SAS (10,000 replications per scenario) evaluates Type I error and power across five parent distributions (Normal, Laplace, t with 3 degrees of freedom, Exponential, and ChiSquare) at sample sizes (10, 10), (20, 10), (20, 20), and (50, 50), under three classes of alternatives: scale-only, location-only, and simultaneous location-and-scale. Empirical and asymptotic correlations between the component statistics and the null sampling distribution of Each combined statistic is reported. Type I error of H+MW is inflated at small samples (0.057). to 0.066 across distributions at (10, 10), reflecting slow convergence to the asymptotic normal distribution rather than correlation between components: the empirical correlation between ZH and ZMW is near zero under H₀, and Type I error approaches the nominal level as sample size grows. Savage+MW maintains Type I error near or below the nominal level across all distributions and sample sizes. Power comparisons at (50, 50) show that no single combined test dominates: H+MW had the broadest performance, leading in 7 of 15 cells with the most balanced power across alternative types; Savage+MW was strongest in several skewed-distribution cells when a location component was present, and WW+MW served as a backup in skewed location-only settings, while AB+MW and ST+MW were not the leading procedures. For a researcher with no prior information about the type of departure, H+MW is recommended as the general-purpose choice; otherwise, the choice can be guided by the expected departure type and the underlying distribution.8 0Item Restricted COLLUSIVE BACKDOOR ATTACKS IN FEDERATED LEARNING: CHALLENGES AND SOLUTIONS(Saudi Digital Library, 2026) Alharbi, Saier; Yifan, GuoThis dissertation work was motivated by the surge in the popularity of federated learning (FL), which is considered as a viable privacy preserving distributed machine learning solution and the security risks associated with its distributed learning architecture. Although FL eliminates the need to store data centrally, it introduces new security risks due to untrustworthy client behavior. In this context, one of the most important and least explored security risks in FL is the collusive backdoor attack (CBA), in which multiple adversaries collude to inject stealthy malicious behavior in the global model. The existing defense strategies, which focus on detecting individual anomalies, do not apply to coordinated attack scenarios, which pose a major security risk to FL systems. The first contribution of this research involved the design and experimental validation of a new framework for a collusive backdoor attack. The proposed perturbation-based attack, which can be conducted collaboratively by adversarial clients, ensures it remains statistically indistinguishable from the behavior of normal clients. The experimental results of the proposed attack, conducted on datasets such as CIFAR-10 and FedEMNIST, showed that it achieves significantly higher attack success rates (ASR) improvements of up to 126% compared to individual backdoor attacks, while maintaining near constant clean input accuracy (CIA ≈ 0.99). The proposed attack was also conducted across scenarios with different levels of malicious participation and varying Gaussian noise levels. The proposed Gram Schmidt-based perturbation estimation ensured that the computational overhead of the proposed attack remained low, with perturbation execution time (≈ 0.1s) remaining constant and negligible compared to the time taken for local training (≈ 5.6s). The second contribution of this research aimed to develop an extensive, well-structured taxonomy of collusive backdoor attacks and countermeasure strategies within FL. While existing surveys on backdoor attacks in FL focus on the broader class, this work proposes an in-depth analysis of collusive backdoor attacks along eight orthogonal dimensions, including optimization methods, coordination methods, temporal execution patterns, aggregation dependencies, knowledge models of adversaries, trigger modalities, etc. This multidimensional taxonomy not only explains how collusive adversaries design and conduct stealthy backdoor attacks but also explains the reasons behind the failures of existing defense strategies, particularly those based on the assumption of independent, statistically deviant malicious clients. By synthesizing results from various studies, the taxonomy reveals some interesting facts about backdoor attacks, including the ability of collusive attackers to maintain high attack success rates while staying within normal statistical bounds, and to evade more than 15 state-of-the-art defense strategies through coordination. The taxonomy also reveals some gaps in the existing research on backdoor attacks, including the lack of collusion-based detection strategies and the limitations of robust aggregation against correlated adversarial updates. Index Terms: Federated learning, Global model, Backdoor attacks, Collusive backdoor attack scheme, Mitigation strategies.15 0Item Restricted OPTIMIZING SMART HOME ENERGY MANAGEMENT SYSTEMS FOR DEMAND RESPONSE USING DEEP REINFORCEMENT LEARNING(2026) Mohammed Sumayli; Olugbenga Moses AnubiThe growing demand for electricity, coupled with the increasing integration of renewable energy sources (RES), has created a critical need for more efficient energy management solutions. Demand Response (DR) plays a vital role in balancing supply and demand, particularly in the residential sector, where flexible energy consumption can help mitigate grid stress and optimize resource use. Home Energy Management Systems (HEMSs) have become central to facilitating residential DR by automating and optimizing household energy usage. Various methods have been employed to optimize HEMSs, including traditional optimization techniques, metaheuristic methods, Model Predictive Control, and robust and stochastic optimization approaches. While these methods offer valuable advantages, they also face significant limitations, such as handling uncertainty, high time complexity, modeling challenges, and limited adaptivity to real-time changes in energy conditions. In contrast, Deep Reinforcement Learning (DRL) has emerged as a promising solution that addresses many of these limitations by offering adaptive, scalable, and real-time decision-making capabilities in dynamic environments. DRL’s ability to learn from interactions with the energy system allows it to handle uncertainties and changing conditions more effectively than traditional approaches. This dissertation further discusses different coordination paradigms for multi-HEMS operation using DRL, the integration of adaptive comfort preferences, and the application of federated learning (FL) as a scalable and privacy-preserving framework for coordinated peak-demand mitigation. Building on this federated formulation, the dissertation also addresses the problem of agent dropout, a practical challenge that arises when households become temporarily unavailable due to communication, computation, or scheduling constraints, and which biases the federated aggregation toward the surviving households. To mitigate this effect, a Confidence-Aware Gradient Substitution (CAGS) algorithm is proposed, combining an exponentially weighted moving average similarity tracker with an adaptive confidence gate that substitutes a dropped agent’s update with that of a trusted neighbour only when the estimated similarity is sufficiently reliable. CAGS is validated on standard federated supervised benchmarks (MNIST, CIFAR-10) and on a heterogeneous federated HEMS deployment, where it consistently outperforms existing dropout-handling baselines, particularly under aggressive dropout ratios.15 0Item Restricted Hedging Oil Price Volatility and US Inflation Surprises: Empirical Evidence from Green Bonds, Precious Metals, Bitcoin, and Passive ETFs(Saudi Digital Library, 2026) AlMajdua, Ibrahim Ah; Cai, LiThis dissertation examines the hedging effectiveness of alternative assets and passive exchange-traded funds against two macro-financial risks, oil price volatility and US inflationary surprises, using dynamic conditional correlation methods. The analysis is organized into two empirical chapters. The first chapter evaluates whether green bonds, precious metals (gold, silver, platinum, palladium), and Bitcoin function as hedges or safe havens against WTI crude oil price fluctuations over August 2014 to August 2024, spanning the 2014-2016 oil price crash, the COVID-19 pandemic, and the Russia-Ukraine conflict. Using a DCC-GARCH framework combined with the Baur and Lucey (2010) hedge/safe-haven classification and variance-based risk reduction metrics, we find that six of eight green bond indices, particularly Credit Aaa, maintain statistically significant negative correlations with oil across normal and stressed market conditions, providing the strongest and most consistent hedging among the three asset classes. Hedging effectiveness is shock-type dependent: green bond correlations strengthen during supply-driven episodes and weaken during demand-driven shocks. Precious metals offer conditional, state-dependent protection, while Bitcoin fails as a reliable hedge and amplifies portfolio variance at weekly horizons. The second chapter evaluates the time-varying inflation-hedging dynamics of 21 tradable passive ETFs against US inflationary surprises from October 2007 to September 2025 using DCC-GARCH estimation, the Baur-Lucey classification adapted to upper-tail inflation risk, and regime-conditional analysis around the March 2022 Federal Reserve tightening cycle. The results establish a four-tier ETF hedge hierarchy: Commodities (mean ρ̄ = +0.15), Sector Equity (+0.08), Broad Equity (+0.05), and Fixed Income (-0.12), with USO leading at +0.33. The iShares TIPS ETF exhibits a duration paradox in which real-rate duration exposure overwhelms CPI indexation during tightening, producing negative inflation correlation precisely when protection is most needed. A two-channel return decomposition formalizes this mechanism. The cross-sectional ranking is persistent across pre- and post-tightening regimes (Spearman rₛ = 0.988), and non-linearity in inflation response is selective rather than universal. Together, the two chapters apply a consistent DCC-GARCH analytical framework to distinguish hedging problems and quantify the conditional, implementable hedging effectiveness available through alternative assets and passive ETFs.7 0Item Restricted OPTIMAL SAMPLING STRATEGIES FOR COMPRESSIVE X-RAY COMPTON BACKSCATTERING IMAGING(Saudi Digital Library, 2026) Alrushud, Abdullah; Gonzalo, ArceCompton backscatter imaging (CBI) is a single-sided X-ray imaging technique in which both the radiation source and the detector are positioned on the same side of the inspected object, enabling non-invasive examination without requiring physical access to both sides. This property makes CBI particularly valuable for security screening, non-destructive testing, and medical diagnostics. Compressive X-ray Compton Backscattering Imaging (CXBI) extends this paradigm by projecting a sequence of coded illumination patterns onto the target, acquiring a reduced set of single-pixel measurements from which the scene is reconstructed computationally. While CXBI has been demonstrated to reduce required radiation dose relative to conventional point-by point scanning, prior work has relied exclusively on random coded apertures. Random patterns, however, are not designed with the sensing matrix properties of CXBI in mind and leave significant room for improvement in both reconstruction quality and radiation efficiency. This dissertation addresses the open problem of coded aperture design for CXBI, establishing that the illumination pattern is not a fixed hardware choice but an optimizable component of the computational imaging system that should be jointly designed with the reconstruction algorithm. Four principal contributions are made. First, a model-based binary aperture design framework is developed, using incoherence, dispersion, binarization, and transmittance regularizers to guide gradient descent optimization of the sensing matrix. Second, a data-driven binary aperture learning framework is introduced that jointly optimizes the sampling mask and a Residual U-block (RSU) based reconstruction network in an end-to-end manner. Third, a physically grounded grayscale CXBI forward model is derived, relating mask pixel thickness to local X-ray transmittance through the Beer-Lambert law and validating five discrete grayscale levels against the Geant-4 Application for Tomographic Emissions (GATE) Monte Carlo simulations. Fourth, three complementary grayscale aperture design strategies are developed: an RSU-based data-driven approach, a joint optimization framework based on the Invertible Diffusion Model (IDM), and a two-stage unrolling mechanism that decouples mask optimization from reconstruction training and uses a mixed signal dependent noise augmentation strategy to improve generalization to realistic GATE acquisition conditions. The proposed frameworks are evaluated through numerical simulations, and the most practically relevant binary, RSU-based, and two-stage grayscale methods are further validated under realistic GATE Monte Carlo conditions, with comparisons against random and model-based coded apertures. In the binary case, the data-driven approach consistently achieves the highest reconstruction quality, outperforming random patterns across all tested scenes. In the grayscale case, the two-stage unrolling framework achieves the best performance under realistic GATE acquisition conditions, recovering finer structural detail than all binary and grayscale baselines and achieving improvements exceeding 1 dB in PSNR and 0.09 in SSIM for multiple test phantoms. Ablation studies confirm that both the optimized mask design and the mixed noise augmentation strategy are essential contributors to GATE generalization, with the combined noise model yielding improvements of 1.62 dB in PSNR and 0.10 in SSIM over training without augmentation. Furthermore, the structured illumination patterns learned by the proposed framework reduce absorbed dose by approximately 26% relative to a random binary baseline, demonstrating that coding optimization simultaneously improves reconstruction fidelity and radiation efficiency. Overall, this dissertation demonstrates that jointly optimized coded apertures, particularly physically realizable grayscale apertures, can improve CXBI reconstruction quality and radiation efficiency relative to random sampling, providing a foundation for future low-dose single-sided X-ray imaging systems.20 0Item Restricted NARRATIVES OF RESISTANCE AND REFORM. EXPLORING GENDER ROLES, SOCIAL JUSTICE AND IDENTITY IN NORTH AND EAST AFRICAN ISLAMIC LITERATURE(Saudi Digital Library, 2026) Alsharif, Budor; Newson, HorstThis dissertation examines African Islamic literature, specifically from North and East Africa, as a dynamic site for ethical inquiry. It majorly focuses on exploration of gender, justice and identity. It also foregrounds African literature written in Arabic, Swahili and Somali as central to the study of Islamic literature. Through a comparative literary methodology, this study analyzes key works, including Tayeb Salih’s Season of Migration to the North, Leila Abouzeid’s Year of the Elephant, Nurrudin Farah’s From a Crooked Rib and Maps, Said Mohammed’s Utengano, and Mahfouz’s The Cairo Trilogy, Ali Mazrui’s The Trial of Christopher Okigbo, alongside Swahili and Somali poems. The dissertation demonstrates that North and East African Islamic Literature actively engages in ethical reinterpretation and critiques moral and political frameworks through fictional works. Drawing on frameworks from the postcolonial theory, African philosophy, and feminist thought, it establishes that concepts such as justice (‘adl), dignity (karama), and community (ummah) are evaluated and reimagined through historical and cultural contexts. Additionally, the dissertation contributes to decolonization of Islamic studies by integrating African ethical systems with Islamic thought and clearly demonstrating that literature can operate as a form of theological reasoning. It concludes that African Islamic Literature offers a vital site for articulation of faith, identity and justice, and in for understanding both regional and global Muslim experiences.11 0Item Restricted Modern Heritage in Riyadh City, Saudi Arabia: The Preservation of Modernist Heritage of the Oil Revolution 1952-1989(Saudi Digital Library, 2026) Khafaji, Hisham; Larkin, Cleary; Baweja, VandanaThis dissertation examines the impact of oil wealth on the built environment of Riyadh, the Saudi Arabian capital, between 1952 and 1989, a period marked by rapid modernization driven by unprecedented petroleum revenues. It focuses on the construction of government institutions that reflect Modernist architectural expressions, representing a departure from vernacular traditions and an engagement with global architectural trends. Growth began in the 1960s but really boomed in the 1970s, with the oil embargo, which saw revenues grow from 2.8 billion to 25 times that amount by 1979. In 1981, the nominal GDP reached 184.3 billion. The expansion ended in 1989 with the first Gulf War. The buildings reviewed include the Ministry of Education (1957), the Institute of Public Administration (1974), the Saudi Central Bank (1978), the Ministry of Foreign Affairs (1984), Saudi Television Headquarters (1982), the General Organization for Social Insurance, main building (1974), the Riyadh Region Office (1987), and the Saudi Fund for Development (1978). By tracing the development of Riyadh's Modernist architecture, the study highlights the socio-political ambitions embedded in the state-led urban transformation and investigates the challenges of preserving this architectural legacy. The dissertation addresses the neglect of Modernist structures, many of which face threats of demolition or redevelopment. Using a multidisciplinary approach that combines archival research, site analysis, and interviews with key players, the dissertation evaluates the historical significance of the key government buildings constructed during the oil boom era. It proposes a heritage assessment framework tailored to the Saudi context, offering guidelines for the preservation and adaptive reuse of Modernist structures. These buildings are architectural artifacts and cultural markers of a pivotal moment in the Kingdom's modernization. Recognizing their heritage value is essential for a more inclusive narrative of Saudi Arabia's architectural history. This study provides a model for reassessing Modernist architecture in other rapidly transforming Gulf countries.2 0Item Restricted LIQUID METAL SODIUM HEAT PIPES FOR NUCLEAR MICROREACTOR APPLICATION, FABRICATION, SYSTEM LEVEL TEST BED OPERATION, AND FILLING RATIO INVESTIGATION(Saudi Digital Library, 2027) Aloufi, Abdulbasit Radhi; Hassan, Yassin A; Ugaz, Victor; Annamalai, Kalyan; Wright, Lesley MThe research is structured through the experimental fabrication process, the accompanying challenges, and the thermal performance evaluation. The first part discusses the fabrication process of sodium heat pipes, including associated safety and impacted performance considerations. A broader investigation is conducted on the wick fabrication, sodium handling and injection procedures, and heat pipe sealing methodologies. These aspects are critically analyzed to establish a reproducible fabrication method that addresses the challenges of high-temperature operation, vacuum integrity, and alkali metal compatibility. The second part is the testing of seven sodium heat pipes integrated with electrically heated fuel rods in a graphite core testbed, representative of a scaled-down design to a single fast-spectrum microreactor core. This facility, developed at the Thermal Hydraulics Laboratory at Texas A&M University, simulates the thermal environment and geometry of compact micro reactor core fuel assembly. The test investigates the operation of steady state conditions and the thermal behavior driven by a thermal graphite core into the coupled bundle of sodium heat pipes. This experiment serves as the conclusive step as it compares active length of the heat pipe with standalone testing for integrating the fabrication methods into a system-level demonstration of passive heat removal in microreactor operation. The final part investigates the performance of the charged amount of the operating fluid, aiming to determine the optimal filling amount for reliable operation. Three heat pipes, identical in geometry and internal wick structure composed of 5 layers stainless-steel mesh (100*100 size), are tested individually under the same experimental conditions. This part compares different sodium filling ratios and examines their influence on startup characteristics, steady-state performance, thermal resistance, and natural convection thermal coefficients. The single pipe testing signifies the figure of merit of thermal response to input power at horizontal orientation.12 0Item Restricted SPORT AS A VEHICLE FOR NATION BRANDING: THE INFLUENCE OF IDENTITY, FANDOM, AND MEDIA ON CULTURE AND TOURISM PERCEPTIONS(Saudi Digital Library, 2026) Barnawi, Mutaz M; Kenonj, A BrownThis dissertation delves into the dynamics of sports-driven nation branding, focusing on identity-based and media-related processes in the United States (U.S.) and Saudi Arabia (KSA) during the pre-event phase of the FIFA World Cup tournament. The present study draws from Social Identity Theory (Tajfel & Turner, 1986) and Anholt's Nation Brand Hexagon (Anholt, 2005a) to investigate how nationalism, patriotism, sport fandom, and social media engagement shape perceptions of a nation's culture and tourism. A quantitative cross-cultural survey design was employed using CloudResearch Prime Panels, resulting in a final sample of 538 respondents (U.S. = 278; KSA = 260). Data were analyzed using exploratory factor analysis, reliability analysis, descriptive statistics, independent-samples t tests, and hierarchical multiple regression. Independent-samples comparisons further indicated that KSA participants reported significantly higher scores than U.S. participants across all seven primary constructs. The findings indicate that nationalism and patriotism are consistent and significant predictors of both culture and tourism perceptions, thereby highlighting the central role of identity-based attachments in shaping national perceptions. Social media engagement significantly predicted perceptions of culture, but showed weaker or inconsistent effects on tourism, whereas social media consumption was not a significant predictor across models. Sport fandom demonstrated a limited role significantly predicting tourism perception in the combined sample; however, it did not predict culture perception or tourism perception within the country specific models. Although the country-specific culture models revealed similar patterns of predictors, the tourism models differed. In the U.S. model, patriotism was the only significant iii predictor, whereas in the KSA model, nationalism, patriotism, and social media engagement were significant predictors, and the model explained more variance. Theoretically, this dissertation extends Social Identity Theory into the domain of nation branding by connecting identity-based constructs to nation-brand outcomes. Furthermore, it facilitates a more targeted implementation of Anholt's nation branding index by illustrating the operation of culture and tourism as distinct, despite their inherent interconnectedness. In practice, the findings provide insights for policymakers and communication strategists seeking to leverage sports and digital media to shape national image prior to mega-events.2 0