Washington University Medical Center
Washington University St. Louis: Open ScholarshipNot a member yet
26344 research outputs found
Sort by
Monitoring and Understanding Momentary Affect Intensity from Smartphone Sensors: Investigating the Role of Time Resolutions and Semantic Locations
No full textMomentary affect intensity refers to the momentary strength or intensity of emotional experience and shows wide associations with critical psychological and behavioral phenomena. Traditional self-report approaches to assess affect intensity require people’s attention and are prone to missingness and recall bias. Wide and cheap smartphone access offers opportunities to monitor and understand real-time affect intensity. This study had three specific aims: (a) To evaluate the potential for a state-of-the-art explainable artificial intelligence (AI) algorithm to predict momentary affect intensities without active inputs, we tested how well Temporal Fusion Transformers (TFTs) can predict momentary affect intensity in an unseen test sample; (b) to explore how model performances and important predictors differ across different time resolutions to aggregated smartphone sensor data, and (c) to explore whether incorporating self-reported semantic location (e.g., home, leisure places) would improve the performances or interpretation of models. We collected data from community adults (N = 179) who completed a 14-day experience sampling method (ESM) protocol of self-reported emotional experiences and continuous monitoring of behaviors through smartphones. The final analytic sample (N = 102) was those who completed at least 40 ESM surveys (out of 70 possible ones). We created a test set (n = 408 surveys) by extracting the last four surveys for each individual, with the rest of the data being the training set (n = 5264 surveys). On the training set, we performed twelve TFTs, with six time resolutions for NA and six for PA, and revealed important predictors in the models. Results showed that in the unseen test set, the best model explained 40.7% of the variance in NA intensity and 38.0% of the variance in PA intensity. The optimal aggregation time periods were three to six hours. Across time resolutions, we found that frequencies of smartphone social interactions (i.e., number of outgoing calls, number of correspondents called) and a mobility marker (i.e., total distance traveled) were consistently identified as top predictors for momentary NA and PA. Incorporating self-report locations was not associated with significant improvement in model performance, but was associated with the pattern of important predictors. Results suggested that TFTs could make moderate predictions of future momentary affect without active input. Aggregating sensor data in three to six hours may be sufficient. Finally, self-reported locations may help refine digital behavioral markers to inform intervention strategies and our understanding of affect-behavior relationships. Overall, the study highlighted the potential of continuous monitoring of affect intensity after a short period of data collection and the potential benefits of explainable AI for both psychological theories and interventions
PI Gain Scheduling and Stall Margin Protection for a JT9D Turbofan Engine Using T-MATS
No full textThe objective of this study was to understand the theory behind turbofan electronic engine controllers (EECs) and to develop a Simulink based controller with stall margin limits, which was then used to build a simple 1-D cruise controller for a Boeing 747. The turbofan model used was the JT9D Pratt & Whitney engine in T-MATS with an attached ”ambient” block to input altitude, Mach, and . Holding altitude at 35000 ft, the controller stall lower and higher limit maps were obtained from the 6% LPC and 15% HPC stall margins at steady state for 0.83-0.85 Mach and 2000-4000 1 RPM. P and I gain schedules were obtained for the same conditions using a first-order model fit through step inputs, as other methods failed to linearize the system. The resulting controller was compared to the example controller given in T-MATS, showing an extended range of operating conditions and stall safety. The system was then fitted to a point-mass 1-D model for a cruise controller, and tested for wind gusts of 73 ft/s . The system showed stability and a settling time of around a minute
MEMS 4110: Charging Connector Extender for Electric Vehicle
No full textThe goal of this project is to develop a mechanism thatwill facilitate automated EV charging, providing supportto disabled individuals. In addition, the system will alsosupport the charging of autonomous vehicles. Thesystem will overcome the weaknesses of wirelesscharging solutions (i.e., uncertain charger alignment,lack of energy efficiency and V2G discharge support). Inthe future, the mechanism could enable fleet chargingof delivery EVs (such as U.S. postal trucks)
MEMS 4110: Egg Project
No full textPeeling hard-boiled eggs by hand is a time-consuming and tedious task that often results in dimples and cracks in the final product. Many chefs and home cooks have tried techniques, from boiling the eggs in baking soda to slipping off egg shells using a spoon, to perfect the peeling process. However, these attempts often fall short by either altering the taste of the egg or extending the amount of time and labor. In the restaurant business, having employees peel eggs for long periods of time is costly. The local St. Louis salad restaurant, Neon Greens, is selling a new protein-dense salad that features a hard-boiled egg on top. As they have been struggling with the egg peeling process, they want a time-efficient solution to the labor-intensive process of peeling hard-boiled eggs
Bargaining Power, Gender Norms, and Intimate Partner Violence: An Examination of Latin American Countries
No full textThis dissertation explores the relationship between women’s employment, bargaining power, gender norms, and intimate partner violence (IPV) in Colombia, the Dominican Republic, Guatemala, and Peru. It is structured around three empirical papers using data from the Demographic and Health Survey, which are nationally representative samples. Paper 1: This paper uses structural equation modeling to examine whether women’s employment and the risk of IPV is mediated through bargaining power and whether that mediation varies across countries. The findings indicate that women’s employment consistently enhances their intrahousehold bargaining power across all countries, highlighting the empowering role of employment. Instead, the path from bargaining power to IPV was nuanced and confounded by control variables such as wealth. Furthermore, the direct path from women’s employment to IPV, or through bargaining power, was minimal. Paper 2: Using generalized linear modeling and marginal effects, this paper investigates whether men’s gender attitudes moderate the relationship between women’s employment status and the probability of IPV and whether the moderation varies by type of IPV or timeframe, using a sample from Colombia. The results show that male partners’ inequitable gender attitudes significantly increase the probability of all types of IPV, irrespective of women’s employment status, with moderation only consistent for economic IPV. Specifically, women experienced higher probability of economic IPV if their spouses held more inequitable gender norms. Paper 3: Employing multilevel modelling analysis, this paper focuses on whether community-level gender norms moderate the relationship between women’s employment and IPV risk, using data from Colombia. The findings show that community norms significantly influence the probability of previous-year physical IPV, with heightened effects among women in cohabiting relationships in settings with intermediate level of unequal gender norms. Altogether, these findings suggest that women’s employment enhances bargaining power, but it does not substantially mediate the relationship with IPV. Instead, male partners’ attitudes and community gender norms play significant roles in influencing the probability of IPV. These results underscore the need for multifaceted economic empowerment strategies coupled with initiatives targeting cultural norms to effectively address IPV and promote gender equality
Metal-Organic Framework-based Biohybrids for Biopreservation and Biosensing
No full textDetection, preservation, and quantification of biomolecules in biological fluids and tissues are fundamental to biomedical research, clinical diagnostics and long-term health monitoring. However, most clinically relevant analytes such as protein biomarkers, antigens and antibody are structurally fragile and exist over wide dynamic ranges in complex matrices such as plasma, serum, and interstitial fluid. Their accurate measurement is often compromised by pre-analytical instability, proteolytic degradation, and the practical constraints of cold-chain storage and centralized laboratory infrastructure. These challenges are more pronounced in minimally invasive sampling routes such as interstitial fluid, where biomarker concentrations are lower and accessible sample volumes are limited. Simple and effective material-based strategies that simultaneously stabilize biomolecules, enrich analytes, and enhance detection sensitivity are therefore essential for advancing early diagnosis, therapeutic monitoring, and disease surveillance, particularly in at home care, rural clinics, and resource limited settings. In this thesis, we introduce a metal organic framework (MOF)-biohybrid platforms that address these challenges through molecular preservation, molecular trapping, and minimally invasive sampling. The first part focuses on zeolitic imidazolate frameworks such as ZIF-8 and ZIF-90 and shows that these materials can encapsulate fragile proteins and entire patient biofluids. Protein biomarkers in plasma, including prostate-specific antigen and panels of cancer-related proteins, as well as SARS-CoV-2 antigens and antibodies used in serologic assays, retain their structural integrity and immunoreactivity after storage at elevated temperatures when protected by these frameworks. Mechanistic design principles are established by quantifying how protein loading density influences ZIF-90 crystal morphology, crystallinity, thermal stability, and preservation efficacy. In the second part, we extend MOF functionality to three dimensional plasmonic biohybrid aerogels, where in situ MOF growth on bacterial nanocellulose and collagen scaffolds, combined with plasmonic nanostructures, yields multifunctional porous materials that analyte pre-concentration with surface enhanced Raman scattering and photothermal antibacterial activity. The final part focuses on microneedle-based devices for in vivo and human interstitial fluid monitoring. Here, conformal MOF coatings on antibody-functionalized microneedles stabilize capture reagents under prolonged thermal stress and ambient transport, while fluorescence amplification restores sensitivity in the face of low analyte concentrations and limited sample volumes. Validated in animal models and human volunteers, these MOF@MN platforms enable quantitative, minimally invasive detection of clinically important protein biomarkers and demonstrate a path toward cold-chain-independent, patient-centric monitoring. Take together, these studies proposed general design principles for MOF-biohybrid systems by relating biomolecule loading, crystal morphology, pore architecture and interfacial engineering to preservation efficacy and sensing performance. By connecting fundamental MOF chemistry with the design of diagnostic assays, three-dimensional sensing materials, environmental monitoring platforms and microneedle devices, it shows how MOF-based biohybrid systems can be engineered to enable environmental monitoring, decentralized diagnostics and continuous health monitoring without strict dependence on the cold-chain
Learning from Images and Text (Clinical) Data: Toward Putting AI in Radiology Workflows
No full textThe rapid growth of medical imaging data highlights the need for intelligent systems that can understand both radiological language and image content. This dissertation presents two complementary research directions that advance machine learning in radiology through large-scale natural language understanding and unsupervised image representation learning. The first part develops and benchmarks machine learning models for identifying follow-up recommendations in radiology reports—a task essential for improving patient care and reducing missed follow-ups. Using 49,769 reports across multiple modalities and three institutional datasets, along with external and temporal test sets, thirty-two classification methods were systematically evaluated on both findings and impression sections. These methods span traditional machine learning, neural networks, and state-of-the-art large language models, including Meta\u27s LLAMA3 and OpenAI\u27s HIPAA-compliant GPT models. Results show that generative-discriminative and attention-based recurrent architecture achieved the best internal performance, while prefixed prompting with GPT-4 offered the strongest external and temporal generalization. This large-scale evaluation establishes a solid foundation for automated extraction of actionable clinical information from radiology reports. The second part introduces Sparse Coding–based Variational Autoencoder (SC-VAE), a new framework for unsupervised image representation learning. SC-VAE integrates sparse coding principles into the VAE architecture through a learnable Iterative Shrinkage-Thresholding Algorithm (ISTA) to enforce sparsity in the latent space. This design addresses key limitations of existing VAEs by learning compact yet expressive representations composed of a small number of orthogonal atoms. Experiments on two image datasets show that SC-VAE achieves superior reconstruction quality compared with state-of-the-art continuous and discrete VAE variants. The learned sparse representations also support effective downstream tasks, including image generation and unsupervised segmentation via patch-level clustering. Together, these studies advance intelligent radiological analysis by improving text understanding and image representation learning. The proposed frameworks demonstrate how scalable machine learning—from clinical text classification to unsupervised generative modeling—may help enhance the interpretability, efficiency, and integration of AI in radiology
Titanium-Cerium Redox Flow Batteries for Grid-Scale Electrical Energy Storage
No full textRedox flow batteries (RFBs) enable large-scale energy storage at low cost due to the independent scaling of device power and energy, thereby unlocking energy arbitrage opportunities and providing a pathway to grid stability and resiliency. Currently, all-vanadium (all-V) RFBs, the only type of RFB which has been commercialized, is facing challenges (e.g., high material cost and low thermodynamic potential) that limit its application. Thus, it is critical to explore the applicability of combination between other redox couples. Furthermore, low energy efficiency (EE) and low capacity (by solubility limitation) are common issues in inorganic RFBs. This dissertation targets on the development of a new RFB system: Titanium (Ti)-Cerium (Ce) RFB. The promising standard potential difference between Ti and Ce couples (1.61 V vs. standard hydrogen electrode, SHE), large abundance of Ti and Ce, and the corresponding cost-effective nature of raw material build up the preconditions for this idea. A highly perm-selective modified poly(ether ketone)-based anion exchange membrane (AEM) is utilized as the separator which ensures long-term operation. The Ti-Ce RFB is tested with both sulfuric acid (H2SO4) and methanesulfonic acid (MSA) supporting electrolytes. In H2SO4, the battery is cycled at 100 mA cm-2 for over 1300 hours with negligible capacity fade and an average EE of 70%. In MSA, the electrolytes exhibit negligible self-discharge after being charged to 90% state of charge (SOC) and stored for 96 hours. After developing the baseline of Ti-Ce RFBs, the second task of this dissertation is performing electrode engineering for MSA-based electrolytes since MSA enables a higher solubility compared to H2SO4 (0.9 vs. 0.5 M). Exploiting the significant difference in reaction kinetics between the Ti and Ce actives, the interfacial area and surface functionalization (affecting electrode-electrolyte contact angles and charge transfer kinetics) of the electrode are optimized to increase operating power while reducing overall cell resistance. An asymmetric electrode configuration which applies carbon paper (CP) and carbon felt (CF) as electrodes of Ce and Ti side, respectively, results in increasing operating current density from 100 to 150 mA cm-2 while sustaining ~70% EE over 80 hours and 100 cycles. The third part of this dissertation is breaking the solubility limit of Ce in traditional acidic supporting electrolytes. Ammonium sulfate (AS) is applied as the supporting electrolyte to enable a supersaturated Ce(IV) electrolyte whose solubility is enhanced to 1.23 M by optimizing the ratio between Ce salt and AS. The study of solution chemistry from characterization techniques and theoretical calculation reveals that the key factor leading to this supersaturated solution is a synergistic effect of Ce(IV) hydrolysis with water and complexation with bisulfate ions (HSO4-). The utilization of AS supporting electrolyte makes Ti-Ce RFBs even more cost-effective, and the EE is stabilized over 70% at 50 mA cm-2