62588 research outputs found
Sort by
Prediction of rTMS treatment outcome in patients with Major Depressive Disorder using EEG and Machine Learning
No full textMajor Depressive Disorder (MDD) remains a significant global health challenge, with profound economic and public health implications. While various treatments do exist, identifying the most effective intervention for individual patients remains troublesome. Failure to achieve treatment response leads to missing the opportunity to improve quality of life for a patient and substantial costs. This study explores the potential of leveraging a neuroimaging technique, electroencephalography (EEG), in combination with advanced machine learning methods to predict treatment outcome in MDD patients undergoing Repetitive Transcranial Magnetic Stimulation (rTMS). Traditional approaches to EEG analysis, such as bandpass filtering followed by power spectrum analysis, suffer from limitations in capturing nuanced neuronal dynamics. To address this, we propose a new machine learning framework incorporating empirical mode decomposition and SBLEST, a Sparse Bayesian learning model. We use iterated masking empirical mode decomposition (itEMD), a data-driven signal processing method for capturing non-stationary oscillations. By decomposing EEG signals into intrinsic mode functions (IMFs), itEMD enables the extraction of both temporal and spectral features crucial for treatment response. Our study integrates itEMD with a novel machine learning model, SBLEST, to predict treatment outcomes. Results demonstrate the superiority of itEMD over traditional bandpass filtering methods, and the effectiveness of SBLEST in predicting treatment response. Additionally, our analysis identifies critical EEG channel features associated with treatment outcomes. While promising, further research with larger sample sizes and broader treatment modalities is warranted to enhance the generalizability and clinical utility of these findings. Overall, this study underscores the potential of machine learning-driven neuroimaging approaches in personalized treatment prediction for MDD, offering hope for improved patient care and outcomes.</p
RHEOLOGICAL CHARACTERIZATION OF COVALENT ADAPTABLE THIOESTER NETWORKS FOR APPLICATIONS IN TISSUE ENGINEERING
No full textCovalent adaptable networks (CANs) are promising platforms for cell delivery. CANs have adaptable cross-links that can rearrange in presence of external stimuli such as pH, shear or light. Due to the ability to rearrange CANs can undergo stress relaxation and possess self-healing properties. This enables them to mimic viscoelastic properties of the ECM.Self-healing properties allow these networks to be delivered by syringe. Different CAN chemistries are being studied for cell delivery. In this work, we characterize thioester CANs designed using 8-arm PEG-thiol and PEG-thioester norbornene. These networks are adaptable due to thioester exchange reaction that can take place in presence of excess thiols in the network. Our goal is to inform design of thioester CANs for cell delivery applications. We use multiple particle tracking microrheology (MPT) and bulk rheology to characterize these scaffolds. MPT is a passive microrheological technique. MPT relies on the thermal motion of probe particles to measure material properties. It is ideal to characterize local changes in microstructure as well as capture phase transitions.
We begin by characterizing thioester scaffolds with different amount of excess thiol. Changing excess thiols changes the networks ability to exchange the cross-links. Scaffolds are made with 0%, 50% and 100% excess thiol. We degrade these networks by incubating them in L-cysteine, which is an amino acid containing a thiol group. L-cysteine will exchange with
network cross-links but has a single thiol, which means it will not participate in network cross-linking. We characterize degradation using multiple particle tracking microrheology (MPT). MPT measures the Brownian motion of fluorescently labeled probe particles embedded in a network. MPT measures rearrangement of each network during degradation.
Networks with 50% excess thiol can only form polymeric clusters during rearrangement while networks with 0% and 100% excess thiol form sample-spanning networks during rearrangement prior to degradation. We then analyze our MPT data with time-cure superposition to calculate the critical relaxation exponent, n, for each network composition. The value
of n is related to the network microstructure. The value of n changes when the amount of excess thiol in the thioester network is changed. Networks with 50% excess thiol are the most elastic networks with n = 0.23 ± 0.04, followed by networks with 0% excess thiol with n = 0.34 ± 0.07. Both these networks are elastic in nature. Networks with 100% excess
thiol have n = 0.53 ± 0.12, which indicates the network is an ideal, percolated network which is equally viscous and elastic in nature. We also measure equilibrium storage moduli of these networks with bulk rheology and a similar trend is measured. Networks with 0%, 50% and 100% excess thiol have moduli of 390 ± 44 Pa, 504 ± 107 Pa and 281 ± 35 Pa,
respectively. Together these results indicate that networks with 50% excess thiol have the highest cross-link density.In this work, we measure decreased cross-link density in thioester networks likely due to network non-idealities including loops, disulfide bond formation and unreacted functional groups. We hypothesize this reduced cross-link density results in the
trend we measure in viscoelastic properties of thioester networks.
We then use PEG thioester scaffolds with 100% excess thiol to 3D encapsulate human mesenchymal stem cells (hMSCs). In this work, we characterize thioester CANs to inform their design as effective cell delivery vehicles. Using bulk rheology, we characterize rearrangement of these networks when they are subjected to strain, which mimics the strain applied by a syringe, and using multiple particle tracking microrheology (MPT) we measure hMSC-mediated remodeling of the pericellular region. Thioester networks are formed by photopolymerizing 8-arm poly(ethylene glycol) (PEG)-thiol and PEG-thioester norbornene. Bulk rheology measures scaffold properties during low and high strain and measures that
thioester scaffolds can recover rheological properties after high strain is applied. We then 3D encapsulate human mesenchymal stem cells (hMSCs) in thioester scaffolds. Using MPT, we characterize degradation in the pericellular region. Encapsulated hMSCs degrade these scaffold within ≈ 4 days post-encapsulation. We hypothesize this degradation is mainly due to cytoskeletal tension that cells apply to the matrix causing adaptable thioester bonds to rearrange leading to degradation. To verify this, we inhibit cytoskeletal tension using blebbistatin, a myosin-II inhibitor. Blebbistatin treated cells can degrade these networks only by secreting enzymes, esterases. Esterases hydrolyze thioester bonds, which generates free thiols leading to bond exchange. Around treated cells, we measure a decrease in the extent of pericellular degradation. We also compare cell area, eccentricity and speed of untreated and treated cells. Inhibiting cytoskeletal tension results in significantly smaller cell area, more rounded cells and lower cell speeds when compared to untreated cells. Overall, this work shows cytoskeletal tension plays a major role in hMSC-mediated degradation of thioester networks. Cytoskeletal tension is also important for spreading and motility of hMSCs in these networks. This work informs the design of thioester scaffolds for tissue regeneration and cell delivery.
We then design thioester networks with enzymatically degradable cross-links as well as adaptable cross-links. Enzy-matically degradable cross-links are formed norbornene functionalized matrix metalloproteinase (MMP)-degradable peptide, KKGPQG↓IWGQKK and adaptable cross-links are formed with PEG-thioester norbornene. We characterize three network compositions with a ratio of 1 : 1, 3 : 1 and 4 : 1 adaptable to MMP-degradable cross-links. We characterize mechanical properties of these networks using bulk rheology. Using MPT, we measure the evolving microstructure of the network during degradation. Our results show that the elastic modulus increases with increasing ratio of adaptable to
MMP-degradable cross-links and all networks have the same extent of stress relaxation. We then measure degradation of these networks by incubating in L-cysteine, which degrades only the adaptable cross-links by the thioester exchange reaction. We measure complete degradation of all three compositions using bulk rheology. Networks with 4 : 1 adaptable
to MMP-degradable cross-links are the slowest to degrade and networks with 3 : 1 adaptable to MMP-degradable cross-links are the fastest to degrade. MPT measurements during degradation show networks with 1 : 1 and 4 : 1 adaptable to MMP-degradable cross-links rearrange multiple times before complete degradation. In networks with 3 : 1 adaptable to
MMP-degradable cross-links, we measure fewer network rearrangements prior to degradation. Using time-cure superposition (TCS), we measure the network structure at the phase transition. Networks with 1 : 1 and 4 : 1 adaptable to MMP-degradable cross-links are elastic and tightly cross-linked and networks with 3 : 1 adaptable to MMP-degradable crosslinks can range from elastic to open networks. The most open network structure, networks with 3 : 1 adaptable to MMP-degradable cross-links, degrades on the shortest timescale. We also measure ≥ 70% hMSC viability in each network after 3D encapsulation. In this work, we characterize different compositions of hybrid networks that incorporate both adaptable and enzymatically degradable cross-links. This work can enable design that specifies the mechanical properties and degradation behavior of the material to better mimic aspects of the native ECM.
Overall, this work characterizes network properties and microstructure of thioester CANs. Measurements of cell- mediated degradation inform about the strategies that hMSCs use to remodel these networks. These studies can enable design of scaffolds that can mimic native cell microenvironment.
</p
Investigation on Surface Oxide Film Fragmentation, Metal Micro-extrusion and Grain Microstructure Evolution During Cold Rolling of Aluminum
No full textThe current investigation aims to understand steps associated with the different stages of solid-state bonding of aluminum alloys as well as the developed grain microstructure within the bond area. The chip-to-chip interactions are studied at a proposed "scaled-up" level using aluminum strips as starting material and cold rolling process as severe plastic deformation condition. Aluminum strips were deliberately covered with anodically grown oxide film at three different thicknesses of 10, 20 and 42µm to tackle the challenge of surface layer observation. To even more precise analysis, anodizing electrolyte is dyed with Indigo carmine for creating colored oxide films. Cold roll process was carried out reduction per passes ranging 2% to 80%. Scanning Electron Microscopy (SEM) equipped with Electron Backscatter Diffraction system (EBSD) was used to characterize the microstructure. Image analysis via Image J and OIM analysis software packages was performed for quantification of the oxide fragment size, crack size, and height of the micro extrusion. Minitab Statistical Software was implemented to develop mathematical models relating the applied strain and surface oxide thickness to the surface oxide fracture trend. A commercial DEFORM-2DTM finite element software package was also used to develop finite element method (FEM) models for evaluation the microextrusion behavior as well as for measuring the local state variables in the solid-state bond area.
The experimental results indicate that the implemented "scale up" method was successful in studying the mechanism of the solid-state bonding of aluminum alloys during rolling process.
It is revealed that the oxide fragment size and the distance between them (crack size) to have decaying and incremental exponential trends respectively by the strain. Regardless of the oxide thickness values, the rate of oxide fragmentation decreased by increasing the strain which leads to higher rate of crack widening. At any reduction per pass, the portion of metal-metal area to oxide-oxide area is independent of the oxide thickness.
Geometric Dynamic Recrystallization (GDRX) is proposed to the phenomenon of emerging the fine equiaxed grains with the bon area. The FEM simulation confirms that at the bond area the value of compression strain reaches to the critical strain value for GDRX happening.
</p
RELIABLE AND STREAMLINED MODEL SETUP FOR DIGITAL TWIN ASSESSMENT OF FRACTURE HEALING
No full textIn large animal preclinical studies of fracture healing, torsional loading is commonly used for postmortem mechanical testing due to its insensitivity to malalignment errors in specimen preparation. Conversely, bending and axial loading, although more physiologically relevant, are highly sensitive to alignment issues in an ex vivo setting due to the natural shape variations and curvature in long bones. Virtual torsional testing using image-based finite element models has been validated in preclinical studies, but its predictive value for capturing whole-bone mechanics and fracture healing quality under other physiologically relevant loading modes has not yet been established. This study aimed to evaluate the association between mechanical biomarkers derived from virtual torsion, axial, and bending tests under strict alignment and malalignment conditions. Computed tomography (CT) scans from 24 intact and operated sheep tibiae and 29 human tibial fractures were used to create digital twins. A method to virtually measure bending flexural rigidity and axial stiffness of the bones models was developed. The results indicated that torsional rigidity is a strong surrogate for bending flexural rigidity in both ovine and human bones. Torsional rigidity and axial stiffness were well-correlated in the ovine data, but only moderately in human fractures due to the complex fracture patterns and bone fragment malalignments. In all models, axial testing was highly prone to error if the applied load and anatomic axis were not perfectly aligned. Based on this study, virtual torsional rigidity is the recommended summary mechanical biomarker of bone healing because it captures variations in healing biomechanics that are present in other loading modes, but the setup is simple and resistant to error from malalignment.</p
Research on Modulation Methods for Underwater Acoustic Communications
No full textThis dissertation explores the advanced modulation schemes for underwater acoustic (UWA) communication through two major contributions: a new Turbo Decision Feedback Equalizer and Decoder (TDFED) for the Orthogonal Time-Frequency Space (OTFS) system, and the Data Dithering Reuse (DDR) method for cross-evaluation of UWA modulation schemes with post-experimental data.The proposed TDFED for the OTFS system employs time-domain feedforward and feedback filters to equalize the received OTFS signals directly without using the delay-Doppler domain approach that is commonly used for OTFS in RF communications. For an OTFS system with Doppler and delay grids, a set of feedforward and feedback equalizers are used in parallel so that the computational complexity is trackable. With low-complexity IPNLMS channel estimation and soft Low-Density Parity Check (LDPC) decoding, the proposed TDFED achieves outstanding performance in heavy Doppler and multipath UWA environments. Extensive lake experiments reveal that the OTFS system significantly reduces the precision requirements of Doppler compensation algorithms, achieving notable improvements in Bit Error Rate (BER) compared to single carrier coherent modulation (SCCM) and orthogonal frequency division multiplexing (OFDM).
The DDR method enables post-experimental cross-evaluation between SCCM and OFDM by reusing the original experimental scheme (OES) data to preserve the nonstationary properties of the UWA channels. This method applies dithering and reverse dithering to transmitted and received data, facilitating the evaluation of new modulation schemes. Field experiments using multiple-input multiple-output (MIMO) measurements demonstrate that the DDR method provides more accurate BER predictions than the existing residual prediction error (RPE)-enhanced simulations.</p
How does changing one health behavior impact another health behavior? Investigating compensation effects
No full textChanging multiple health behaviors at once can be more effective in reducing the risk of chronic disease compared to changing a single health behavior alone; however, few theories address the theoretical mechanisms of multiple health behavior change. The Compensatory Carry-over Action Model (CCAM) is one such theory that explains how one behavior interacts with another behavior, but few studies have tested the hypotheses in this model. This thesis will examine one mechanism by which behaviors interact: compensatory cognitions (CCs). Previous research on CCs has mostly focused on the negative impacts that CCs have on behavior change, however, different types of CCs may impact intentions differently, such that some CCs may lead to stronger behavioral intentions. Moreover, perceived instrumentality may play an important moderating role in the relationship between CCs and intentions. This study of college students (N = 162) demonstrates that perceived instrumentality of a behavior is positively associated with intentions for that behavior. Additionally, we found a negative association between endorsement of the belief that eating fruits and vegetables can compensate for a lack of exercise with exercise intentions. No evidence was found for any positive associations among CCs and intentions nor for perceived instrumentality as a moderator of these relationships. These results could be used to help inform multiple health behavior change interventions. </p
The Recent Historiography on Chinese Immigration in the 19th-Century United States
No full textThis historiographical essay provides an overview of recent literature on Chinese immigration in the 19th-century United States. Its first part sketches how recent American, Asian American, and Chinese historians investigate the story of Chinese railroad workers and gold seekers using varied primary sources and distinguished perspectives. The second part examines the last three decades of scholars who researched Chinese prostitution in the 19th century America West and how they used Chinese female immigration as a perspective to narrate the gender and racial history of the U.S. The third part traces scholars\u27 opinions about the U.S. legal framework for immigrants. The essay argues that Chinese immigration illuminates intersections between global migration and international relations and illustrates the U.S. modern history. The evolution in recent literature on Chinese immigration reveals the rise of micro-histories as a counter to the larger time scale of global history.</p
Examining the Relationship between Reading and Math: Within-Year Growth on Star Assessments
No full textThe purpose of this study is to model the interrelationship of reading and mathematics as measured by commonly used computer adaptive tests (CATs), Star Reading (SR) and Star Math (SM), for students within a single school year. Reading and math show theoretical and empirical evidence for being interconnected skills, following a similar developmental trajectory of basic skills to higher order reasoning. However, prior research has primarily focused on reading and math as separate skills, studied primarily elementary grades, and examined individually administered assessments. The current study is the first investigation of the relationship as measured by CATs. Participants were students in grades 1, 4, 8, and 11 who took SR and SM in a single school year. Bivariate latent growth modeling was used to assess model fit for proposed reading to math, math to reading, and bidirectional reading to math relationships. Best fitting models indicated that grade 1 showed a math to reading relationship, whereas grades 4, 8, and 11 showed a bidirectional relationship. Structural model differences were observed. Results are discussed in the context of theory, prior empirical research, and assessment. Limitations, implications for practice, and future directions are discussed
Mechanistic Modelling of Electrochemical Processes on Gold using Electrochemical Impedance Spectroscopy
No full textThis dissertation focuses on the mechanistic modeling of electrochemical reactions that occur on gold surfaces including anodic gold oxide formation in acidic and basic solutions, potential dependent adsorption of bisulfate, sulfate, and oxygen on gold in sulfuric acid solution, and the hydrogen evolution reaction in sulfuric acid solutions using electrochemical impedance spectroscopy. Despite years of studies carried out on the electrochemistry of gold including oxide formation, electrochemical behavior of gold in the double layer region in the presence of adsorbed ions, and the hydrogen-evolution reaction (HER) mechanism in acidic and basic solutions, there is still disagreement regarding the mechanisms of the processes. To address this problem, we studied the electrochemistry of gold in acidic and basic solutions by developing mechanistic impedance-based models, which offer a clearer picture of electrochemical processes on gold. Chapters 3 and 4 describe the anodic formation of gold oxide in sulfuric acid and potassium hydroxide solutions using the framework of the Point Defect Model and electrochemical impedance spectroscopy. Chapter 5 describes the potential dependent adsorption of bisulfate, sulfate, and oxygen in sulfuric acid solution, using the Anion Catalyzed Active Dissolution Model. Even though many studies have been carried out on the hydrogen-evolution reaction mechanism in acidic and alkaline solutions, most used dc-polarization techniques to measure exchange current density and Tafel slopes, though considerably less attention has been given to the kinetics of the HER. Therefore, for the HER in sulfuric acid solutions, a mechanistic impedance-based model based on Volmer-Heyrovsky-Tafel mechanism was developed. The models were then validated against experimental observations. These studies on the electrochemistry of gold provide unique insights and information on the kinetics of oxide formation in acidic and basic solutions, the behavior of gold in the double layer and pre-oxide formation regions in sulfuric acid solutions as well as kinetics of the HER in acidic solutions