Search

INTRODUCTION: To reduce demands on expert time and improve clinical efficiency, we developed a framework to evaluate whether inexpensive, accessible data could accurately classify Alzheimers disease (AD) clinical diagnosis and predict the likelihood of progression. METHODS: We stratified relevant data into three tiers: obtainable at primary care (low-cost), mostly available at specialty visits (medium-cost), and research-only (high-cost). We trained several machine learning models, including a hierarchical model, an ensemble model, and a clustering model, to distinguish between diagnoses of cognitively unimpaired, mild cognitive impairment, and dementia due to AD. RESULTS: All models showed viable classification, but the hierarchical and ensemble models outperformed the conventional model. Classifier error was predictive of progression rates, and cluster membership identified subgroups with high and low risk of progression within 1.5 to 3 years. DISCUSSION: Accessible, inexpensive clinical data can be used to guide AD diagnosis and are predictive of current and future disease states. HIGHLIGHTS: Classification performance using cost-effective features was accurate and robustHierarchical classification outperformed conventional multinomial classificationClassification labels indicated significant changes in conversion risk at follow-upA clustering-classification method identified subgroups at high risk of decline.

The capability to adapt to dynamic environments and changes in data distribution is essential for the development of adaptive artificial intelligence (AI) systems that can operate effectively in the real world. Typically, adaptive learning requires an AI agent to autonomously gather the required information (i.e., training data) from the deployed environment to adapt to the local dynamic changes. This process falls within the broad area of semi-supervised learning, where partial training data is not manually labeled. This is especially critical for applications at the edge of cyber-physical systems, where available datasets are rather limited and resources for deep learning from big data are not available. While there exist effective semi-supervised learning methods based on feature smoothness assumptions, they become less robust in dynamic environments where there are shifts (e.g., concept drift) in data distribution. Recently, causality has been explored to address deployment domain shifts. This is based on the argument that causality, especially the causal direction, is consistent across different domains, making invariant features highly significant in adaptive learning. This dissertation presents an interactive causality (IC) methodology, which utilizes directional and temporal causal events to facilitate the automated self-labeling of data. Interactive causality refers to the temporal state transitions of causes and effects during interactive activities, which are captured through additional observing channels. The methodology capitalizes on the domain-invariant causal relationships to capture information leakage during interactions from an additional observing channel, and is fortified by the rich domain knowledge in cyber-physical system contexts. Differing from big data ML approaches, the IC leverages the existing knowledge and experiences to create a knowledge graph with embedded temporal causality among interactive nodes and directly look at the interacti

Policy learning from other international contexts is an important strategy during the sport policy making process for the government of Taiwan, and recent research has examined the potential parallels between Taiwan and Scotland with regards to sports policy. Although the status of Taiwan and Scotland is not the same, interesting comparisons can be made given their shared status as ‘contested nations’ that are often in the shadow of their closest neighbours with whom there is an uneasy political relationship – respectively, China and England. As a consequence, sport is regarded in both countries as an important vehicle for establishing and promoting a distinctive identity, albeit with contrasting political and policy considerations. Drawing upon 15 semi-structured interviews with sports policymakers and politicians from both the Taiwanese and Scottish contexts, this paper critically examines the similarities and contrasts with regards to the political considerations which shape and constrain the nature of sport policy in each context. This analysis will focus on the role of central government, local government, sport policy organisations, and sporting National Governing Bodies in both Taiwan and Scotland, with particular emphasis on the positioning of sport within the broader policy, political, ideological and constitutional considerations for policymakers in each context.

The jet-in-crossflow is a critical flowfield for a range of aerospace propulsion systems, both airbreathing and rocket engines. A jet-in-crossflow or transverse jet is typically composed of a jet issuing from a circular outlet perpendicularly to a uniform air crossflow. Previous experiments deploying hot-wire anememetry or laser-based optical diagnostics such as planar laser-induced fluorescence (PLIF) or particle image velocimetry (PIV) have found that there is a transition in the jet's upstream shear layer from convective to absolute instability as the jet-to-crossflow velocity ratio $R$ or momentum flux ratio $J$ decreases. Typically, convectively unstable jets-in-crossflow are correlated with more asymmetric cross-sections while more symmetric jet cross-sections are typically associated with absolutely unstable transverse jets at lower $R$ values, usually below $3$ (Gevorkyan et al., 2016). Besnard et al. (2022) demonstrated that even low-amplitude asymmetric acoustic excitation affected the symmetry of a convectively unstable jet-in-crossflow. Such a finding suggests that there may be a linear origin to symmetry-breaking in the face of a theoretically symmetric flow field. This dissertation describes a numerical investigation of the spatial linear stability characteristics of a jet-in-crossflow for the axisymmetric and helical azimuthal modes using a 2D-local viscous linear stability analysis. This study uses the viscous hyperbolic-tangent (tanh) and uniformly valid asymptotic solution (UVAS) base flows developed by Alves et al. (2008) and Alves & Kelly (2008) as viscous extensions of the cylindrical vortex sheet solution of Coelho & Hunt (1989), who performed asymptotic expansion-based linear stability analyses of the same base flows. This present work may be considered a `fully-coupled' extension of their work because we presently account for all base flow-eigenfunction azimuthal coupling.The eigenspectra from the present fully-coupled linear stabilit

Diagnostic classification models (DCMs) have been used to classify examinees into groups based on their possession status of a set of latent traits. In addition to traditional item-based scoring approaches, examinees may be scored based on their completion of a series of small and similar tasks. Those scores are usually considered as count variables. To model count scores, this study proposes a new class of DCMs that uses the negative binomial distribution at its core. We explained the proposed model framework and demonstrated its use through an operational example. Simulation studies were conducted to evaluate the performance of the proposed model and compare it with the Poisson-based DCM.

Modern statistical and machine learning models excel at capturing complex non-linear relationships between outcomes and predictors, resulting in high accuracy. However, the complexity of these models can impede statistical inference and interpretation. This dissertation confronts and tries to overcome the emerging challenges presented by intricate models and big data.One significant challenge involves modeling and statistical inference for zero-inflated semi-continuous data. Thus, in the first part, we develop a flexible Bayesian semi-parametric mixture model for zero-inflated skewed longitudinal data, generating credible intervals for not only the mean but also any quantiles of the parameters and predictions, aiding population inference of skewed data. The model is applied to evaluate how number of binge drinking episodes changes with neuromaturation using the National Consortium on Alcohol and Neuro-Development in Adolescence data.On the other hand, credible or confidence intervals do not directly address a common question: can we identify a subset of predictions or parameters with true values exceeding a specific threshold with confidence? To tackle this, in the second part, we improve upon the inverse set estimation framework that estimates such sets by developing an approach with fewer assumptions and broader applicability to various data settings. We construct an excursion set map with probability guarantee on the North American Regional Climate Change Assessment Program data using the proposed method. Moreover, we use this new method to discover characteristics of in-patients at high risk for severe outcomes using University of California San Diego hospital data.In the third part, we apply this inverse set estimation inference framework to quantify prediction model uncertainty and develop theories and algorithms that ensure non-conservative coverage rates for a single threshold in non-asymptotic settings in regression problems. We demonstrate the effectivenes

Fragile X syndrome (FXS), the most prevalent heritable form of intellectual disability, is caused by the transcriptional silencing of the FMR1 gene. While neuronal contribution to FXS has been extensively studied in both animal and human-based models of FXS, the roles of astrocytes, a type of glial cells in the brain, are largely unknown. Here, we generated a human-based FXS model via differentiation of astrocytes from human-induced pluripotent stem cells (hiPSCs) and human embryonic stem cells (hESCs) and characterized their development, function, and proteomic profiles. We identified shortened cell cycle, enhanced Ca2+ signaling, impaired sterol biosynthesis, and pervasive alterations in the proteome of FXS astrocytes. Our work identified astrocytic impairments that could contribute to the pathogenesis of FXS and highlight astrocytes as a novel therapeutic target for FXS treatment.

Allelic tagging of endogenous genes enables studying gene function and transcriptional control in the native genomic context. Here, we present an efficient protocol for bi-allelic tagging of protein-coding genes with fluorescent reporters in human iPSCs using the CRISPR-Cas9-mediated homology-directed repair. We detail steps for design, cloning, electroporation, and single-cell clone isolation and validation. The tagging strategy described in this protocol is readily applicable for knockin of other reporters in diverse cell types for biomedical research.

The gut is continuously invaded by diverse bacteria from the diet and the environment, yet microbiome composition is relatively stable over time for host species ranging from mammals to insects, suggesting host-specific factors may selectively maintain key species of bacteria. To investigate host specificity, we used gnotobiotic Drosophila, microbial pulse-chase protocols, and microscopy to investigate the stability of different strains of bacteria in the fly gut. We show that a host-constructed physical niche in the foregut selectively binds bacteria with strain-level specificity, stabilizing their colonization. Primary colonizers saturate the niche and exclude secondary colonizers of the same strain, but initial colonization by Lactobacillus species physically remodels the niche through production of a glycan-rich secretion to favor secondary colonization by unrelated commensals in the Acetobacter genus. Our results provide a mechanistic framework for understanding the establishment and stability of a multi-species intestinal microbiome.

Cost effectiveness is one of the crucial key factors driving the shipping and trading industry. Nowadays global trade involves multiple parties and sectors across international borders, and numerous paper transport documents, such as paper bills of lading (pBLs), are handled and disseminated around the world. With the advent of digital technology, electronic solutions are arguably ripe for the replacement of pBLs. The key benefits of using an electronic bill of lading have long been identified as the speed of transfer, the avoidance of external costs and removal of need to use Letters of Indemnity. As an alternative to the transfer of physical endorsements, electronic bills of lading (eBLs) are processed and transferred automatically by a computer system that constructs the information electronically using agreed standards. However, the journey of promoting eBLs is not all smooth sailing. Despite the growing number of private and public projects and initiatives during the past 40 years, the maritime world has yet seen the dawn of the electronic age. On the contrary, pBLs remain a staple in international trade, despite years of criticism over delays, costs and security risks. This thesis seeks to identify the underlying reasons for the stagnation in the development of this key legal document. It will demonstrate that, apart from the technical challenges, the real impediment to a true replacement of the pBL is the formidable legal challenges in reproducing the three principal functions of the antique paper-based trade document, for which the existing paperless implementations have failed to provide anything but a piecemeal solution. In coming to this finding, the thesis will also examine the effectiveness of the various legislative efforts, both nationally and internationally, to address the legal issues identified in the thesis. Finally, the author will attempt to speculate on the trends and prospects of the evolution of eBLs in the light of the observation

Cellular process is meticulously regulated by means of transcription and additional epigenetic mechanisms. This regulation is further complicated by the communication between cells, which coordinate gene expression across multiple cells and tissues. Epigenetic modifications such as DNA methylation, histone modification can induce heritable changes in gene expression without alterations to the DNA. Dysregulation of epigenome or transcriptome often leads to various diseases, such as cancer, developmental disorders, and neurodegenerative diseases. Recent advances in single-cell RNA sequencing (scRNA-seq) and spatially resolved sequencing has provided us unprecedented insights into cellular heterogeneity, developmental trajectories, spatial organization, and cell-cell communications. However, how the cells are spatially and temporally regulated through transcriptome, epigenome, and intercellular communication, is still not clearly understood. In this dissertation, we studied this issue from three distinct perspectives. First, we investigated the maintenance of DNA methylation throughout the cell cycle. Specifically, by analyzing experimental data, we found the post-replication methylation maintenance rates are correlated between nearby CpGs in a region-specific manner. Through stochastic modeling, we derived evidences for genome-wide methyltransferase processivity in cells, and developed an approach to infer lengthscales of linear diffusion of DNA-binding proteins using the rate correlation. In the second project, we developed a deep learning method to identify the spatiotemporal organization of cells. We further proposed the pseudo-Spatiotemporal Map (pSM) as a spatial counterpart of the pseudotime in scRNA-seq. We validated the accuracy of the pSM using public datasets and demonstrated its usefulness in revealing developmental sequences and providing insights into cancer progression. In the last project, we developed a method for summarizing the major cell-cell commun

RouteNav is an iOS app designed to support wayfinding for blind travelers in an indoor/outdoor transit hub. It doesn’t rely on ex- ternal infrastructure (such as BLE beacons); instead, localization is obtained by fusing spatial information from inertial dead reck- oning and GPS (when available) via particle filtering. Routes are expressed as sequences of “tiles”, where each tile may contain rele- vant points of interest. Redundant modalities are used to guide users to switching goalposts within tiles. In this paper, we describe the different components of RouteNav, and report on a user study with seven blind participants, who traversed three challenging routes in a transit hub while receiving input from the app.

Although structures of vitrified supramolecular complexes have been determined at near-atomic resolution, elucidating in situ molecular structure in living cells remains a major challenge. Here, we apply a novel but simple liquid-cell technique, developed previously for real-time imaging of the dynamics at a liquid-gas interface, to image wet biological samples. With extra scattering from the liquid phase, the transmission electron micrographs show amplitude contrast comparable to that in negatively stained samples. Single-molecule domains are resolved in the protein complex GroEL imaged in buffer solution at room temperature. Moreover, various stages of virus cell entry, which are transient events with very few structural information to date, are also captured. Morphological details are reconstructed using the technique of individual particle electron tomography. These results demonstrate that this approach can be a valuable yet cost-effective technique complementary to other microscopy techniques for addressing important biological questions at the molecular level.

The United States is currently experiencing its worst drug crisis, which is largely driven by opioid addiction and primarily due to fentanyl. It is therefore necessary to investigate the mechanisms mediating fentanyl's rewarding and reinforcing properties to contribute to the development of successful treatment strategies. Gut bacteria communicate with the brain, and vice versa, via the gut-brain axis to regulate brain function, mood, and behavior. Addiction is a chronic brain disorder that alters circuitry involved in reward, stress, learning, and motivation, all of which have a bidirectional influence between their associated behaviors and gut bacteria. Given the associations between opioid use, gastrointestinal distress, and microbial dysbiosis in humans and rodents, I tested the hypothesis that interactions between gut bacteria and the brain mediate the reinforcing and motivational properties of fentanyl. In this dissertation, I present my work that supports a bidirectional relationship between gut bacteria and fentanyl intravenous self-administration (IVSA) in Sprague Dawley rats. In the following studies, I implanted rats with intravenous catheters in preparation for fentanyl IVSA on an escalating schedule of reinforcement and analyzed gut microbiota by sequencing bacterial DNA from rat fecal samples. I demonstrate that based on sex and fentanyl dose, the diversity of gut bacteria is either increased or decreased following fentanyl IVSA and predicts progressive ratio breakpoint, a measure of motivation. Further, I show that depletion of gut bacteria via prolonged oral antibiotic treatment enhances fentanyl IVSA, and restoration of microbial metabolites with short-chain fatty acid administration decreases fentanyl IVSA back to controls at higher fixed ratio schedules of reinforcement. My findings highlight an important relationship between the knockdown and rescue of gut bacterial metabolites and fentanyl self-administration in adult rats, which provides suppor

BACKGROUND & AIMS: There is an ongoing debate on whether non-alcoholic fatty liver disease (NAFLD) is an active contributor or an innocent bystander in the pathogenesis of coronary artery disease (CAD). The aim of the present study was to assess the causal relationship between NAFLD and CAD.APPROACH & RESULTS: We performed two-sample Mendelian randomization (MR) analyses using summary-level data to assess the association between genetically predicted NAFLD (i.e. chronically-elevated serum alanine aminotransferase levels [cALT], imaging-based and biopsy-confirmed NAFLD) and risk of CAD. Analyses were repeated after exclusion of NAFLD susceptibility genes that are associated with impaired VLDL secretion. Inverse-variance weighted (IVW) MR analyses showed a statistically significant association between genetically predicted cALT and risk of CAD (odds ratio [OR]:1.116, 95% confidence interval [CI]:1.039,1.199), but not for the other NAFLD-related traits (OR:1.046, 95%CI:0.764,1.433 and OR:1.014, 95%CI:0.968,1.062 for imaging-based and biopsy-confirmed NAFLD, respectively). MR Egger regression revealed a statistically significant intercept, indicative of directional pleiotropy, for all traits. Repeat analyses after exclusion of genes associated with impaired VLDL secretion, showed consistent associations between genetically predicted NAFLD and CAD for all traits, i.e. cALT (OR:1.203, 95%CI:1.113,1.300), imaging-based (OR:2.149, 95%CI:1.276,3.620) and biopsy-confirmed NAFLD (OR:1.113, 95%CI:1.041,1.189), which persisted when more stringent biopsy-confirmed NAFLD criteria were used (OR:1.154, 95%CI:1.043,1.278) or when more stringent MR methods were applied. MR Egger regression did not show a statistically significant intercept.CONCLUSION: The two-sample MR analyses showed a robust association between genetically predicted NAFLD and CAD after exclusion of genetic variants that are implicated in impaired VLDL secretion.

Assessments with a large amount of small, similar, or often repetitive tasks are being used in educational, neurocognitive, and psychological contexts. For example, respondents are asked to recognize numbers or letters from a large pool of those and the number of correct answers is a count variable. In 1960, George Rasch developed the Rasch Poisson counts model (RPCM) to handle that type of assessment. This article extends the RPCM into the world of diagnostic classification models (DCMs) where a Poisson distribution is applied to traditional DCMs. A framework of Poisson DCMs is proposed and demonstrated through an operational dataset. This study aims to be exploratory with recommendations for future research given in the end.

We report innovative scalable, vertical, ultra-sharp nanowire arrays that are individually addressable to enable long-term, native recordings of intracellular potentials. Stable amplitudes of intracellular potentials from 3D tissue-like networks of neurons and cardiomyocytes are obtained. Individual electrical addressability is necessary for high-fidelity intracellular electrophysiological recordings. This study paves the way toward predictive, high-throughput, and low-cost electrophysiological drug screening platforms.

PurposeTo improve the performance of less experienced clinicians in the diagnosis of benign and malignant spinal fracture on MRI, we applied the ResNet50 algorithm to develop a decision support system.MethodsA total of 190 patients, 50 with malignant and 140 with benign fractures, were studied. The visual diagnosis was made by one senior MSK radiologist, one fourth-year resident, and one first-year resident. The MSK radiologist also gave the binary score for 15 qualitative imaging features. Deep learning was implemented using ResNet50, using one abnormal spinal segment selected from each patient as input. The T1W and T2W images of the lesion slice and its two neighboring slices were considered. The diagnostic performance was evaluated using tenfold cross-validation.ResultsThe overall reading accuracy was 98, 96, and 66% for the senior MSK radiologist, fourth-year resident, and first-year resident, respectively. Of the 15 imaging features, 10 showed a significant difference between benign and malignant groups with p < = 0.001. The accuracy achieved by using the ResNet50 deep learning model for the identified abnormal vertebral segment was 92%. Compared to the first-year resident's reading, the model improved the sensitivity from 78 to 94% (p < 0.001) and the specificity from 61 to 91% (p < 0.001).ConclusionOur deep learning-based model may provide information to assist less experienced clinicians in the diagnosis of spinal fractures on MRI. Other findings away from the vertebral body need to be considered to improve the model, and further investigation is required to generalize our findings to real-world settings.

Smoking remains the leading cause of preventable death in the United States. Although combustible cigarettes are largely being replaced by tobacco-free products, nicotine use continues to increase in vulnerable populations, including youth, adolescents, and pregnant women. Nicotine exerts unique effects on specific brain regions during distinct developmental periods due to the dynamic expression of nicotinic acetylcholine receptors (nAChRs) throughout the lifespan. Nicotine exposure is a health concern not only for adults but also has neurotoxic effects on the fetus, newborn, child, and adolescent. In this review, we aim to highlight the dynamic roles of nAChRs throughout gestation, adolescence, and adulthood. We also provide clinical and preclinical evidence of the neurodevelopmental, cognitive, and behavioral consequences of nicotine exposure at different developmental periods. This comprehensive review highlights unique effects of nicotine throughout the lifespan to help elucidate interventions and public health measures to protect sensitive populations from nicotine exposure.

Natural killer (NK) cells are a major subset of innate immune cells that are essential for host defense against pathogens and cancer. Two main classes of inhibitory NK receptors (NKR), KIR and CD94/NKG2A, play a key role in suppressing NK activity upon engagement with tumor cells or virus-infected cells, limiting their antitumor and antiviral activities. Here, we find that single-chain NKR antagonists linked to a VHH that binds the cell surface phosphatase CD45 potentiate NK and T activities to a greater extent than NKR blocking antibodies alone in vitro. We also uncovered crosstalk between NKG2A and Ly49 that collectively inhibit NK cell activation, such that CD45-NKG2A and CD45-Ly49 bispecific molecules show synergistic effects in their ability to enhance NK cell activation. The basis of the activity enhancement by CD45 ligation may reflect greater antagonism of inhibitory signaling from engagement of MHC I on target cells, combined with other mechanisms, including avidity effects, tonic signaling, antagonism of weak inhibition from engagement of MHC I on non-target cells, and possible CD45 segregation within the NK cell-target cell synapse. These results uncover a strategy for enhancing the activity of NK and T cells that may improve cancer immunotherapies.

Communication between the brain and gut bacteria impacts drug- and addiction-related behaviors. To investigate the role of gut microbiota on fentanyl reinforcement and reward, we depleted gut bacteria in adult Sprague Dawley male and female rats using an oral, nonabsorbable antibiotic cocktail and allowed rats to intravenously self-administer fentanyl on an escalating schedule of reinforcement. We found that antibiotic treatment enhanced fentanyl self-administration in males, but not females, at the lowest schedule of reinforcement (i.e., fixed ratio 1). Both males and females treated with antibiotics self-administered greater amounts of fentanyl at higher schedules of reinforcement. We then replete microbial metabolites via short-chain fatty acid administration to evaluate a potential mechanism in gut-brain communication and found that restoring metabolites decreases fentanyl self-administration back to controls at higher fixed ratio schedules of reinforcement. Our findings highlight an important relationship between the knockdown and rescue of gut bacterial metabolites and fentanyl self-administration in adult rats, which provides support for a significant relationship between the gut microbiome and opioid use. Further work in this field may lead to effective, targeted treatment interventions in opioid-related disorders.

Semi-continuous data present challenges in both model fitting and interpretation. Parametric distributions may be inappropriate for extreme long right tails of the data. Mean effects of covariates, susceptible to extreme values, may fail to capture relevant information for most of the sample. We propose a two-component semi-parametric Bayesian mixture model, with the discrete component captured by a probability mass (typically at zero) and the continuous component of the density modeled by a mixture of B-spline densities that can be flexibly fit to any data distribution. The model includes random effects of subjects to allow for application to longitudinal data. We specify prior distributions on parameters and perform model inference using a Markov chain Monte Carlo (MCMC) Gibbs-sampling algorithm programmed in R. Statistical inference can be made for multiple quantiles of the covariate effects simultaneously providing a comprehensive view. Various MCMC sampling techniques are used to facilitate convergence. We demonstrate the performance and the interpretability of the model via simulations and analyses on the National Consortium on Alcohol and Neurodevelopment in Adolescence study (NCANDA) data on alcohol binge drinking.

Inhibitory neurons (INs) consist of distinct subtypes with unique functions. Previous studies on INs mainly focused on single brain regions, and thus it remains unclear whether the modulation of IN subtypes occurs globally across multiple regions. Here, we monitored the activity of different cortical IN subtypes at both macroscale and microscale in mice learning a lever-press task. Learning evoked a global modulation of IN subtypes throughout the cortex. The initial learning phase involved strong activation of vasoactive intestinal peptide-expressing INs (VIP-INs) and weak activation of somatostatin-expressing INs (SOM-INs). Inactivating VIP-INs increased SOM-IN activity and impaired initial learning. Concurrently, cortical cholinergic inputs from the basal forebrain were initially more active but became less engaged over learning. Manipulation of the cholinergic system impaired motor learning and differentially altered activity of IN subtypes. These results reveal that motor learning involves a global and subtype-specific modulation on cortical INs regulated by the cholinergic system.

Purpose The impact of the COVID-19 pandemic on crime has been highly variable. One possible source of variation runs indirectly through the impact that the pandemic had on groups tasked with preventing and responding to crime. Here, this paper aims to examine the impact of the pandemic on the activities undertaken by front-line workers in the City of Los Angeles Mayor’s Office of Gang Reduction and Youth Development (GRYD). Design/methodology/approach The authors use both autoregressive integrated moving average modeling and a regression-based event study design to identify changes in GRYD Community Intervention Worker proactive peacemaking and violence interruption activities induced by the onset of the City of Los Angeles “safter-at-home” lockdown. Findings Analyses show that the proactive peacemaking and violence interruption activities either remained stable or increased with the onset of the lockdown. Originality/value While the City of Los Angeles exempted GRYD’s Community Intervention Workers from lockdown restrictions, there was no guarantee that proactive peacemaking and violence interruption activities would continue unchanged. The authors conclude that these vital functions were indeed resilient in the face of major disruptions to daily life presented by the pandemic. However, the causal connection between stability in Community Intervention Worker activities and gang-related crime remains to be evaluated.

Big data plays an important role in the fourth industrial revolution, which requires engineersand computers to fully utilize data to make smart decisions to optimize industrial processes. In the additive manufacturing (AM) industry, laser powder bed fusion (LPBF) and direct metal laser solidification (DMLS) have been receiving increasing interest in research because of their outstanding performance in producing mechanical parts with ultra-high precision and variable geometries. However, due to the thermal and mechanical complexity of these processes, printing failures are often encountered, resulting in defective parts and even destructive damage to the printing platform. For example, heating anomalies can result in thermal and mechanical stress on the building part and eventually lead to physical problems such as keyholing and lack of fusion. Many of the aforementioned process errors occur during the layer-to-layer printing process, which makes in-situ process monitoring and quality control extremely important. Although in-situ sensors are extensively developed to investigate and record information from the real-time printing process, the lack of efficient in-situ defect detection techniques specialized for AM processes makes real-time process monitoring and data analysis extremely difficult. Therefore, to help process engineers analyze sensor information and efficiently filter monitoring data for transport and storage, machine learning and data processing algorithms are often implemented. These algorithms integrate the functionality of automated data processing, transferring, and analytics. In particular, sensor data often takes the form of images, and thus, a prominent approach to conducting image analytics is through the use of convolutional neural networks (CNN). Nevertheless, the industrial utilization of machine learning methods often encounters problems such as limited and biased training datasets. Hence, simulation methods, such as the finite-element method

Our understanding of the ocean historically has moved forward in parallel with our ability to make observations. In the thesis, high-resolution observations of the California Current System made by Spray underwater gliders are used to discuss extreme events, eddy across-shore transport, and the annual cycle of dissolved oxygen in the upper ocean. The time scales covered in the thesis include annual to interannual changes while the spatial scales are mesoscale and larger. The availability of high-resolution ocean glider data for over 13 years provides the backbone to conduct analyses over these time and spatial scales. The thesis starts by examining temperature and salinity extremes from 2014-2019 in the California Current System and its source waters. The 2014-2019 period was anomalously warm. In addition, a salinity anomaly from 2017-2019 in the California Current System was found to have formed in the North Pacific Subtropical Gyre in 2015 and subsequently advected into the source waters of the California Current. Next, the thesis examines the offshore propagation of subthermocline eddies from the coast. Subthermocline eddies are observed to propagate at near the local first baroclinic Rossby wave speed. It is estimated that the subthermocline eddies are important to the salt budget in the California Current System and are difficult to track with surface observations alone. The thesis next discusses dissolved oxygen observations collected from 2017 to 2020. First, the thesis considers the procedure to correct for drift in the optical sensors used to make dissolved oxygen observations. A model is fit to changes in the gain correction coefficient over time and predicts the drift for 5 years after sensor calibration. Second, the thesis describes the annual cycle of dissolved oxygen in the upper 500 m of the central and southern California Current System. A subsurface dissolved oxygen maximum is described in the oligotrophic region on the offshore edge of the Californ

The memory system has been evolving at a fast pace recently, driven by the emergence of large-scale applications and the advance in hardware technology. This trend calls for the birth of big memory systems with extreme heterogeneity, which combines multiple memory technologies with different latency, bandwidth, and capacity to construct main memory.The conventional memory management methods are not adequate to handle the increasing complexity that heterogeneity brings forward, and will fail to deliver the full potential of the new memory. Specifically, the heterogeneity of memory systems brings a substantial disparity in the performance and efficiency, making the decision of which technology to use at what times intricate.The performance of heterogeneous memory-based big memory systems highly depends on the data locality.By dynamically profiling memory access behaviors, memory management solutions move frequently accessed (or hot) data from slow to fast memory and less frequently accessed (or cold) data from fast to slow memory. However, the decision of how to spread data across all memory components in big memory systems, which data to move, and at what times is not trivial. This dissertation proposes a series of techniques, spanning from architecture, runtime systems, operating systems, programming models, to applications and algorithms, which are used to efficiently identify which data in applications are performance-critical, timely determine when to perform data migration, and select where to place those data in big memory systems. Specifically, this dissertation contains six common scenarios using big memory systems with big memory applications such as machine learning/artificial intelligence applications, large-scale scientific simulation, and in-memory databases. We propose several software techniques for the efficient use of big memory systems in all scenarios. By doing so, this dissertation identifies bottlenecks in the existing memory management solution

Gut bacteria influence neural circuits in addiction-related behaviors. Given the association between opioid use, gastrointestinal distress, and microbial dysbiosis in humans and mice, we test the hypothesis that interactions between gut bacteria and the brain mediate the rewarding and reinforcing properties of fentanyl. We implant rats with intravenous catheters in preparation for fentanyl intravenous self-administration (IVSA) on an escalating schedule of reinforcement to determine factors that influence fentanyl intake, including sex, dose, and gut microbiota. Our data show the impact of fentanyl IVSA on gut microbiota diversity, as well as the role of gut microbiota on fentanyl IVSA, in Sprague Dawley rats in a sex- and dose-dependent manner (n = 10-16/group). We found that the diversity of gut microbiota within females dose-dependently predicts progressive but not fixed ratio schedules of fentanyl IVSA. Depending on sex and fentanyl dose, alpha diversity (richness and evenness measured with Shannon index) is either increased or decreased following fentanyl IVSA and predicts progressive ratio breakpoint. Our findings collectively suggest a role of gut bacteria in drug-related behavior, including motivation and reinforcement. This work provides feasibility for an intravenous fentanyl self-administration model and uncovers potential factors mediating drug use, which may lead to the development of effective addiction interventions.

When developing ordinal rating scales, we may include potentially unordered response options such as "Neither Agree nor Disagree," "Neutral," "Don't Know," "No Opinion," or "Hard to Say." To handle responses to a mixture of ordered and unordered options, Huggins-Manley et al. (2018) proposed a class of semi-ordered models under the unidimensional item response theory framework. This study extends the concept of semi-ordered models into the area of diagnostic classification models. Specifically, we propose a flexible framework of semi-ordered DCMs that accommodates most earlier DCMs and allows for analyzing the relationship between those potentially unordered responses and the measured traits. Results from an operational study and two simulation studies show that the proposed framework can incorporate both ordered and non-ordered responses into the estimation of the latent traits and thus provide useful information about both the items and the respondents.

Protein secondary structure discrimination is crucial for understanding their biological function. It is not generally possible to invert spectroscopic data to yield the structure. We present a machine learning protocol which uses two-dimensional UV (2DUV) spectra as pattern recognition descriptors, aiming at automated protein secondary structure determination from spectroscopic features. Accurate secondary structure recognition is obtained for homologous (97%) and nonhomologous (91%) protein segments, randomly selected from simulated model datasets. The advantage of 2DUV descriptors over one-dimensional linear absorption and circular dichroism spectra lies in the cross-peak information that reflects interactions between local regions of the protein. Thanks to their ultrafast (∼200 fs) nature, 2DUV measurements can be used in the future to probe conformational variations in the course of protein dynamics.

Objectives Dispersion of bone dust in the posterior fossa during retrosigmoid craniectomy for vestibular schwannoma (VS) resection could be a source of meningeal irritation and lead to development of persistent postoperative headaches (POH). We aim to determine risk factors, including whether the presence of bone spicules that influence POH after retrosigmoid VS resection. Design Present study is a retrospective case series. Setting The study was conducted at a tertiary skull-base referral center. Participants Adult patients undergoing VS resection via a retrosigmoid approach between November 2017 and February 2020 were included for this study. Main Outcome Measures Development of POH lasting ≥ 3 months is the primary outcome of this study. Results Of 64 patients undergoing surgery, 49 had complete data (mean age, 49 years; 53% female). Mean follow-up time was 2.4 years. At latest follow up, 16 (33%) had no headaches, 14 (29%) experienced headaches lasting <3 months, 19 (39%) reported POH lasting ≥3 months. Twenty-seven (55%) patients had posterior fossa bone spicules detectable on postoperative computed tomography (CT). Age, gender, body mass index, length of stay, tumor diameter, size of craniectomy, the presence of bone spicules, or the amount of posterior petrous temporal bone removed from drilling did not differ significantly between patients with POH and those without. On multivariate logistic regression, patients with POH were less likely to have preoperative brainstem compression by the tumor (odds ratio [OR] = 0.21, p = 0.028) and more likely to have higher opioid requirements during hospitalization (OR = 1.023, p = 0.045). Conclusion The presence of bone spicules in the posterior fossa on postoperative CT did not contribute to headaches following retrosigmoid craniectomy approach for VS resection.

One major challenge in analyzing spatial transcriptomic datasets is to simultaneously incorporate the cell transcriptome similarity and their spatial locations. Here, we introduce SpaceFlow, which generates spatially-consistent low-dimensional embeddings by incorporating both expression similarity and spatial information using spatially regularized deep graph networks. Based on the embedding, we introduce a pseudo-Spatiotemporal Map that integrates the pseudotime concept with spatial locations of the cells to unravel spatiotemporal patterns of cells. By comparing with multiple existing methods on several spatial transcriptomic datasets at both spot and single-cell resolutions, SpaceFlow is shown to produce a robust domain segmentation and identify biologically meaningful spatiotemporal patterns. Applications of SpaceFlow reveal evolving lineage in heart developmental data and tumor-immune interactions in human breast cancer data. Our study provides a flexible deep learning framework to incorporate spatiotemporal information in analyzing spatial transcriptomic data.

The sinoatrial node (SAN), the leading pacemaker region, generates electrical impulses that propagate throughout the heart. SAN dysfunction with bradyarrhythmia is well documented in heart failure (HF). However, the underlying mechanisms are not completely understood. Mitochondria are critical to cellular processes that determine the life or death of the cell. The release of Ca2+ from the ryanodine receptors 2 (RyR2) on the sarcoplasmic reticulum (SR) at mitochondria-SR microdomains serves as the critical communication to match energy production to meet metabolic demands. Therefore, we tested the hypothesis that alterations in the mitochondria-SR connectomics contribute to SAN dysfunction in HF. We took advantage of a mouse model of chronic pressure overload-induced HF by transverse aortic constriction (TAC) and a SAN-specific CRISPR-Cas9-mediated knockdown of mitofusin-2 (Mfn2), the mitochondria-SR tethering GTPase protein. TAC mice exhibited impaired cardiac function with HF, cardiac fibrosis, and profound SAN dysfunction. Ultrastructural imaging using electron microscope (EM) tomography revealed abnormal mitochondrial structure with increased mitochondria-SR distance. The expression of Mfn2 was significantly down-regulated and showed reduced colocalization with RyR2 in HF SAN cells. Indeed, SAN-specific Mfn2 knockdown led to alterations in the mitochondria-SR microdomains and SAN dysfunction. Finally, disruptions in the mitochondria-SR microdomains resulted in abnormal mitochondrial Ca2+ handling, alterations in localized protein kinase A (PKA) activity, and impaired mitochondrial function in HF SAN cells. The current study provides insights into the role of mitochondria-SR microdomains in SAN automaticity and possible therapeutic targets for SAN dysfunction in HF patients.

Understanding the heterogeneity and inefficiencies of human pluripotent stem cell (hPSC) differentiation are key to robust human tissue design. This thesis combines live multi-day, multi-colour high content fluorescence microscopy movies of hPSCs stably co-expressing fluorescently tagged histone H2B and the two-colour FUCCI cell cycle reporter with quantitative image analysis to inform both top-down and bottom-up mathematical approaches of studying hPSC behaviour and cell fate decisions. Using CHASTE, an agent based cell modelling toolkit, I implemented in silico agent-based simulations to model hPSC proliferation, colony formation and migration in 2D informed by experimentally observed cell cycle and motility parameters derived from manually quantitated fluorescent hPSC timelapse image sequences. I developed a computational image processing pipeline to quantitatively characterise hPSC populations and dynamics and to obtain high-dimensional quantitative single-cell level feature information from ’live’ hPSCs in order to inform top-down modelling of hPSC behaviour. This pipeline captures several hundred morphological, proliferation and fluorescence associated features for hundreds of thousands of cells across different experimental conditions for downstream analysis and prediction of cell fate. Using the image analysis tools we developed, I helped quantitatively characterise ORACLE, a novel class of nuclear-rim localised fluorescent cell fate reporters recently developed in our group. Using that pipeline I morphologically profiled hPSC cell populations during pluripotency maintenance as well as during early differentiation toward early germ layer (ectoderm, mesoderm, endoderm) cell fates. Coupling this approach with deep learning approaches allowed us to achieve accurate and robust cell fate phenotyping ’live’. We were able to demonstrate that this cell phenotyping method is capable of predicting cell fate changes as early as fluorescent transcription factor based ce

The increase of ocean plastic in the Gal ́apagos Archipelago receives more and more attention in recent years. However, the sources and the potential processes that cause the locally found plastic remain unknown. In previous research, the emerged plastic on Gal ́apagos coastlines is proved to be more likely from external sources. Among all the land-based plastic sources, riverine plastic inputs to the ocean are considered the main sources. In this research, we attribute the Gal ́apagos plastic to the major riverine sources on the west coastlines of America based on a Bayesian inference framework. Rivers in Panama Bight and Surrounding Regions are the most likely sources. We also found that most of the particles from the Panama Bight and Surrounding Regions only arrive in the Gal ́apagos from January to April. Particles from North Humboldt Current System also show an episodic nature of arrivals, but the arrival time is more spread throughout the years. Using a high-resolution ocean surface currents model, we simulate the pathways of virtual floating plastic particles from the river sources to the Gal ́apagos region. We analyzed the pathways and potential ocean dynamics that cause the episodic arrivals. A Panama Jet Current is first defined in this research and is considered to be responsible for most of the plastic arriving in the Gal ́apagos from the Panama Bight and Surrounding Regions. Humboldt Current is expected to transport most of the particles from North Humboldt Current System sources to the Gal ́apagos. Attributing the riverine sources and identifying the ocean transporting dynamics help to better understand the episodic nature of arriving particles in the Gal ́apagos Archipelago. Eventually, we are one step closer to predicting the arrival time of plastic in the Gal ́apagos Archipelago.

This paper seeks to investigate whether the impact of credit risk and liquidity risk on the profitability of Chinese commercial banks has changed as a result of the covid-19. To this end, we adopted financial data and indicators for a total of 32 Chinese listed commercial banks from 2009 to 2021 and conducted an empirical analysis using a fixed-effects model. Our empirical findings suggested that covid-19 strengthened the negative impact of credit risk on profitability, with the most significant effect on city commercial banks and a significant effect on ROA only for joint-stock commercial banks, while it has no significant effect on state-controlled large commercial banks. We also find that covid-19 weakened the positive effect of liquidity risk on profitability, but it has a significant effect on ROA only for joint-stock commercial banks. This study provides evidence of the impact of the epidemic on commercial banks, enabling them to manage risk well and improve their risk tolerance in the post-epidemic era, as well as providing empirical evidence for future research.

Understanding the heterogeneity and inefficiencies of human pluripotent stem cell (hPSC) differentiation are key to robust human tissue design. This thesis combines live multi-day, multi-colour high content fluorescence microscopy movies of hPSCs stably co-expressing fluorescently tagged histone H2B and the two-colour FUCCI cell cycle reporter with quantitative image analysis to inform both top-down and bottom-up mathematical approaches of studying hPSC behaviour and cell fate decisions. Using CHASTE, an agent based cell modelling toolkit, I implemented in silico agent-based simulations to model hPSC proliferation, colony formation and migration in 2D informed by experimentally observed cell cycle and motility parameters derived from manually quantitated fluorescent hPSC timelapse image sequences. I developed a computational image processing pipeline to quantitatively characterise hPSC populations and dynamics and to obtain high-dimensional quantitative single-cell level feature information from ’live’ hPSCs in order to inform top-down modelling of hPSC behaviour. This pipeline captures several hundred morphological, proliferation and fluorescence associated features for hundreds of thousands of cells across different experimental conditions for downstream analysis and prediction of cell fate. Using the image analysis tools we developed, I helped quantitatively characterise ORACLE, a novel class of nuclear-rim localised fluorescent cell fate reporters recently developed in our group. Using that pipeline I morphologically profiled hPSC cell populations during pluripotency maintenance as well as during early differentiation toward early germ layer (ectoderm, mesoderm, endoderm) cell fates. Coupling this approach with deep learning approaches allowed us to achieve accurate and robust cell fate phenotyping ’live’. We were able to demonstrate that this cell phenotyping method is capable of predicting cell fate changes as early as fluorescent transcription factor based ce

The increase of ocean plastic in the Gal ́apagos Archipelago receives more and more attention in recent years. However, the sources and the potential processes that cause the locally found plastic remain unknown. In previous research, the emerged plastic on Gal ́apagos coastlines is proved to be more likely from external sources. Among all the land-based plastic sources, riverine plastic inputs to the ocean are considered the main sources. In this research, we attribute the Gal ́apagos plastic to the major riverine sources on the west coastlines of America based on a Bayesian inference framework. Rivers in Panama Bight and Surrounding Regions are the most likely sources. We also found that most of the particles from the Panama Bight and Surrounding Regions only arrive in the Gal ́apagos from January to April. Particles from North Humboldt Current System also show an episodic nature of arrivals, but the arrival time is more spread throughout the years. Using a high-resolution ocean surface currents model, we simulate the pathways of virtual floating plastic particles from the river sources to the Gal ́apagos region. We analyzed the pathways and potential ocean dynamics that cause the episodic arrivals. A Panama Jet Current is first defined in this research and is considered to be responsible for most of the plastic arriving in the Gal ́apagos from the Panama Bight and Surrounding Regions. Humboldt Current is expected to transport most of the particles from North Humboldt Current System sources to the Gal ́apagos. Attributing the riverine sources and identifying the ocean transporting dynamics help to better understand the episodic nature of arriving particles in the Gal ́apagos Archipelago. Eventually, we are one step closer to predicting the arrival time of plastic in the Gal ́apagos Archipelago.

This paper seeks to investigate whether the impact of credit risk and liquidity risk on the profitability of Chinese commercial banks has changed as a result of the covid-19. To this end, we adopted financial data and indicators for a total of 32 Chinese listed commercial banks from 2009 to 2021 and conducted an empirical analysis using a fixed-effects model. Our empirical findings suggested that covid-19 strengthened the negative impact of credit risk on profitability, with the most significant effect on city commercial banks and a significant effect on ROA only for joint-stock commercial banks, while it has no significant effect on state-controlled large commercial banks. We also find that covid-19 weakened the positive effect of liquidity risk on profitability, but it has a significant effect on ROA only for joint-stock commercial banks. This study provides evidence of the impact of the epidemic on commercial banks, enabling them to manage risk well and improve their risk tolerance in the post-epidemic era, as well as providing empirical evidence for future research.

Protein secondary structure discrimination is crucial for understanding their biological function. It is not generally possible to invert spectroscopic data to yield the structure. We present a machine learning protocol which uses two-dimensional UV (2DUV) spectra as pattern recognition descriptors, aiming at automated protein secondary structure determination from spectroscopic features. Accurate secondary structure recognition is obtained for homologous (97%) and nonhomologous (91%) protein segments, randomly selected from simulated model datasets. The advantage of 2DUV descriptors over one-dimensional linear absorption and circular dichroism spectra lies in the cross-peak information that reflects interactions between local regions of the protein. Thanks to their ultrafast (∼200 fs) nature, 2DUV measurements can be used in the future to probe conformational variations in the course of protein dynamics.

One major challenge in analyzing spatial transcriptomic datasets is to simultaneously incorporate the cell transcriptome similarity and their spatial locations. Here, we introduce SpaceFlow, which generates spatially-consistent low-dimensional embeddings by incorporating both expression similarity and spatial information using spatially regularized deep graph networks. Based on the embedding, we introduce a pseudo-Spatiotemporal Map that integrates the pseudotime concept with spatial locations of the cells to unravel spatiotemporal patterns of cells. By comparing with multiple existing methods on several spatial transcriptomic datasets at both spot and single-cell resolutions, SpaceFlow is shown to produce a robust domain segmentation and identify biologically meaningful spatiotemporal patterns. Applications of SpaceFlow reveal evolving lineage in heart developmental data and tumor-immune interactions in human breast cancer data. Our study provides a flexible deep learning framework to incorporate spatiotemporal information in analyzing spatial transcriptomic data.

Gut bacteria influence neural circuits in addiction-related behaviors. Given the association between opioid use, gastrointestinal distress, and microbial dysbiosis in humans and mice, we test the hypothesis that interactions between gut bacteria and the brain mediate the rewarding and reinforcing properties of fentanyl. We implant rats with intravenous catheters in preparation for fentanyl intravenous self-administration (IVSA) on an escalating schedule of reinforcement to determine factors that influence fentanyl intake, including sex, dose, and gut microbiota. Our data show the impact of fentanyl IVSA on gut microbiota diversity, as well as the role of gut microbiota on fentanyl IVSA, in Sprague Dawley rats in a sex- and dose-dependent manner (n = 10-16/group). We found that the diversity of gut microbiota within females dose-dependently predicts progressive but not fixed ratio schedules of fentanyl IVSA. Depending on sex and fentanyl dose, alpha diversity (richness and evenness measured with Shannon index) is either increased or decreased following fentanyl IVSA and predicts progressive ratio breakpoint. Our findings collectively suggest a role of gut bacteria in drug-related behavior, including motivation and reinforcement. This work provides feasibility for an intravenous fentanyl self-administration model and uncovers potential factors mediating drug use, which may lead to the development of effective addiction interventions.

Big data plays an important role in the fourth industrial revolution, which requires engineersand computers to fully utilize data to make smart decisions to optimize industrial processes. In the additive manufacturing (AM) industry, laser powder bed fusion (LPBF) and direct metal laser solidification (DMLS) have been receiving increasing interest in research because of their outstanding performance in producing mechanical parts with ultra-high precision and variable geometries. However, due to the thermal and mechanical complexity of these processes, printing failures are often encountered, resulting in defective parts and even destructive damage to the printing platform. For example, heating anomalies can result in thermal and mechanical stress on the building part and eventually lead to physical problems such as keyholing and lack of fusion. Many of the aforementioned process errors occur during the layer-to-layer printing process, which makes in-situ process monitoring and quality control extremely important. Although in-situ sensors are extensively developed to investigate and record information from the real-time printing process, the lack of efficient in-situ defect detection techniques specialized for AM processes makes real-time process monitoring and data analysis extremely difficult. Therefore, to help process engineers analyze sensor information and efficiently filter monitoring data for transport and storage, machine learning and data processing algorithms are often implemented. These algorithms integrate the functionality of automated data processing, transferring, and analytics. In particular, sensor data often takes the form of images, and thus, a prominent approach to conducting image analytics is through the use of convolutional neural networks (CNN). Nevertheless, the industrial utilization of machine learning methods often encounters problems such as limited and biased training datasets. Hence, simulation methods, such as the finite-element method

Our understanding of the ocean historically has moved forward in parallel with our ability to make observations. In the thesis, high-resolution observations of the California Current System made by Spray underwater gliders are used to discuss extreme events, eddy across-shore transport, and the annual cycle of dissolved oxygen in the upper ocean. The time scales covered in the thesis include annual to interannual changes while the spatial scales are mesoscale and larger. The availability of high-resolution ocean glider data for over 13 years provides the backbone to conduct analyses over these time and spatial scales. The thesis starts by examining temperature and salinity extremes from 2014-2019 in the California Current System and its source waters. The 2014-2019 period was anomalously warm. In addition, a salinity anomaly from 2017-2019 in the California Current System was found to have formed in the North Pacific Subtropical Gyre in 2015 and subsequently advected into the source waters of the California Current. Next, the thesis examines the offshore propagation of subthermocline eddies from the coast. Subthermocline eddies are observed to propagate at near the local first baroclinic Rossby wave speed. It is estimated that the subthermocline eddies are important to the salt budget in the California Current System and are difficult to track with surface observations alone. The thesis next discusses dissolved oxygen observations collected from 2017 to 2020. First, the thesis considers the procedure to correct for drift in the optical sensors used to make dissolved oxygen observations. A model is fit to changes in the gain correction coefficient over time and predicts the drift for 5 years after sensor calibration. Second, the thesis describes the annual cycle of dissolved oxygen in the upper 500 m of the central and southern California Current System. A subsurface dissolved oxygen maximum is described in the oligotrophic region on the offshore edge of the Californ

The memory system has been evolving at a fast pace recently, driven by the emergence of large-scale applications and the advance in hardware technology. This trend calls for the birth of big memory systems with extreme heterogeneity, which combines multiple memory technologies with different latency, bandwidth, and capacity to construct main memory.The conventional memory management methods are not adequate to handle the increasing complexity that heterogeneity brings forward, and will fail to deliver the full potential of the new memory. Specifically, the heterogeneity of memory systems brings a substantial disparity in the performance and efficiency, making the decision of which technology to use at what times intricate.The performance of heterogeneous memory-based big memory systems highly depends on the data locality.By dynamically profiling memory access behaviors, memory management solutions move frequently accessed (or hot) data from slow to fast memory and less frequently accessed (or cold) data from fast to slow memory. However, the decision of how to spread data across all memory components in big memory systems, which data to move, and at what times is not trivial. This dissertation proposes a series of techniques, spanning from architecture, runtime systems, operating systems, programming models, to applications and algorithms, which are used to efficiently identify which data in applications are performance-critical, timely determine when to perform data migration, and select where to place those data in big memory systems. Specifically, this dissertation contains six common scenarios using big memory systems with big memory applications such as machine learning/artificial intelligence applications, large-scale scientific simulation, and in-memory databases. We propose several software techniques for the efficient use of big memory systems in all scenarios. By doing so, this dissertation identifies bottlenecks in the existing memory management solution