Mason Journals (George Mason Univ.)
Not a member yet
3256 research outputs found
Sort by
Familial Eosinophilic Esophagitis: A Systems-Level Analysis of Immune and Epithelial Transcriptional Programs
No full textEosinophilic esophagitis (EoE) is a chronic, immune-mediated disorder marked by esophagealeosinophilic infiltration and a Th2-skewed inflammatory profile. Familial aggregation, with asevenfold increased risk among first-degree relatives, suggests a strong genetic component,though the molecular mechanisms linking inherited risk to disease onset remain unclear.
To investigate this, we analyzed RNA sequencing data (GSE250595) from esophageal biopsiesof 68 individuals: 43 EoE patients, 6 unaffected relatives, and 19 unrelated healthy controls.Differential expression analysis (DESeq2), Gene Ontology (GO) enrichment, and WeightedGene Co-expression Network Analysis (WGCNA) were applied to uncover pathways and genemodules associated with disease status.
Our results revealed immune activation and extracellular matrix remodeling as dominant featuresin affected individuals, whereas unaffected relatives showed enrichment for antigen presentationand immune surveillance pathways. WGCNA identified three key modules: (1) the MEbluemodule—positively correlated with healthy controls—was enriched for epithelial maintenanceand stress-response genes (e.g., HSPA5, HIF1A); (2) the MEgreen module, linked to disease,included immune signaling genes (CD4, FN1) and three EoE-associated rare variants (MUC16,ADGRE1, TENM3); and (3) the MEturquoise module, enriched for proliferative and survivalpathways involving TP53 and AKT1. Notably, zinc ion response genes—including MT1X,MT1E, and MT2A—were downregulated in patients and clustered within the MEblue module,indicating impaired epithelial stress adaptation and barrier repair.
These findings suggest that familial EoE involves loss of protective transcriptional programsalongside activation of inflammatory and proliferative signaling. Our integrative analysishighlights zinc dysregulation and immune overactivation as central to disease progression,offering mechanistic insight and potential avenues for early intervention in geneticallypredisposed individuals
A Bioinformatics Approach to Identifying Differentially Expressed Genes in Multiple Sclerosis Subtypes
No full textMultiple sclerosis (MS) is a chronic autoimmune disorder of the central nervous system characterized by the loss of myelin (demyelination), which disrupts nerve signal transmission and can lead to lasting neurological damage. Clinically, MS presents with a range of symptoms such as numbness, muscle weakness, impaired coordination, and visual disturbances, and it manifests in several forms including primary progressive (PPMS), relapsing-remitting (RRMS), and secondary progressive MS (SPMS).
In this study, we analyzed gene expression profiles from 16 individuals (including PPMS, RRMS, SPMS patients and healthy controls) to identify genes differentially expressed in each MS subtype. Several genes exhibited subtype-specific dysregulation. Notably, RRMS samples showed strong upregulation of IFNA14 and two long intergenic non-coding RNAs (LINC01994 and LINC01995), along with significant downregulation of LINC01478 and NEUROD1, whereas SPMS samples were marked by reduced expression of PKP4.
IFNA14 encodes a type I interferon cytokine involved in lymphocyte activation during immune responses. NEUROD1 is a neurogenic transcription factor that regulates insulin gene expression and contributes to neuronal differentiation and repair. PKP4 (plakophilin-4) is a junctional plaque protein involved in organizing cell–cell junctions and cadherin-mediated adhesion. The differential regulation of these genes highlights potential molecular markers and pathways underlying MS progression, suggesting new targets for further investigation
Multispectral RGB-LWIR Fusion with YOLO for Autonomous Object Detection in Low-Cost Mobile Robotics under Variable Lighting Conditions
No full textMultispectral imaging (MSI), combining RGB and long-wave infrared (LWIR) data, has shown promise for improving object detection in challenging and dynamic lighting conditions, such as low-light or high-glare outdoor environments. While RGB-based computer vision is commonly used, it often fails in these situations, whereas thermal imaging, though robust in poor lighting, sacrifices fine detail. Despite this, there is a lack of practical lightweight systems for evaluating how fused MSI data performs in edge-computing scenarios, particularly on mobile robotics platforms. This project addresses that gap by assessing the effectiveness of RGB, LWIR, and fused imagery for human detection using YOLOv5, YOLOv8, and YOLOv11 architectures. A mecanum mobile robot equipped with RGB and LWIR cameras captures multispectral data under varying lighting conditions. Images are sent to a FastAPI web app to perform spatial image registration, fuse the pictures with the SeAFusion algorithm, and use lightweight custom YOLO models to return information on detected humans. With these outputs, the robot uses PID control and ultrasonic sensing to follow a person. We trained nine YOLO models (three architectures across three sensor modes) on a single human class and evaluated performance quantitatively through mAP and precision/recall metrics, as well as qualitatively through real-world tracking trials. Results indicate that multispectral fusion improves detection robustness, as the fused model achieved a 4.9% higher [email protected] and 6.3% increase in precision compared to thermal-only inputs. This research shows the potential of low-cost MSI fusion and processing for real-time applications in robotics and edge AI
Enabling Multi-Node Underwater Image Transmission with a LoRa Mesh Network
No full textUnderwater communication is essential for applications such as marine research, environmental monitoring, and infrastructure inspection. However, traditional methods like acoustic, Radio Frequency signals, and optical signaling are often expensive, power-intensive, and can negatively impact marine life–particularly due to water’s permeability. To address these challenges, this study details a low-cost, low-power underwater communication system built on a multi-hop LoRa mesh network. The system utilizes Seeeduino LoRaWAN boards with RHF76-052AM modules, submerged in polycarbonate containers and powered by Lithium Polymer batteries. A mesh topology on the application layer routes byte-converted data packets across multiple submerged nodes, which are then reassembled at the final receiver, with upwards of 3 nodes. System performance was validated in a controlled underwater environment by testing packet transmission across multiple Spreading Factors (SF7-SF12), with signal quality and link strength quantified using the Received Signal Strength Indicator (RSSI) metric. These results confirm the viability of LoRa for multi-node underwater networks, establishing it as an effective, scalable, and eco-friendly alternative for low-bandwidth data transmission in aquatic environments
No-verdose: Using SEIRP and Optimal Control Theory to Model the Opioid Epidemic and Mitigate Opioid-Induced Mortality
No full textThe opioid epidemic was declared a public health emergency in 2017, and continues to worsen the wellbeing of communities and account for hundreds of thousands of deaths annually. Drug epidemics can be modeled using Susceptible-Infected-Recovered (SIR) epidemiological models, which simulate the dynamics of addiction and inform intervention optimization strategies. In this project, we model opioid addiction using a variation of the SIR model called the SEIRP model that includes Exposed (E) and Prevented (P) compartments. We then introduce a control variable representing a digital education and prevention campaign as an epidemic intervention. We implement optimal control theory using Pontryagin’s Maximum Principle and the Forward-Backward Sweep Algorithm to maximize the impact of the education campaign. We use literate programming to enhance the communication of our research process. The uniqueness of our SEIRP model is that it accounts for individuals who avoid opioids for a lifetime due to prevention efforts. We found that for both quadratic and linear cost functions, the optimal intervention intensity remained at its maximum value at the beginning of the time frame. For quadratic cost, the optimal intervention intensity decreased continuously from the middle to the end of the time frame. For linear cost, the optimal intervention intensity dropped discontinuously to its minimum value in the middle and remained at its minimum through the end of the time frame. This project informs future opioid epidemic mitigation policies, in particular the effectiveness of digital education campaigns in relation to the costs of these efforts. Our research contributes to the United Nations’ Sustainable Development Goal #3: Good Health and Well-Being
Evaluating the Performance of Three Flood Impact Analysis Tools
No full textFlood modeling is a critical part of disaster preparedness and mitigation planning, especially in coastal and riverine urban areas prone to storm surge and flash flooding. Accurately predicting flood impacts can help both civilians and government officials prepare and reduce risk in the event of a flooding event. However, there are many free flood analysis tools that vary in effectiveness and accuracy of analysis. One option to determine risk is to look at flood maps to identify areas within flood zones; but this approach does not account for the depth of potential flooding, damage to infrastructure, or other quantifiable metrics. Other solutions, such as HAZUS-MH and FEMA FAST offer a more detailed analysis by considering the cost of building replacement, and the foundation materials used for construction. Elevation, building data, and a simulated flood mirroring flooding impacts of Hurricane Isobel for the city of Baltimore, MD was consistent throughout analysis. Both HAZUS and FAST yielded overestimates, with FAST being more accurate by 83%. Although there is no clear best tool for mitigation planning, FEMA FAST should be the preferred choice for flood risk analysis
Enhancing Low-Cost Sensor Reliability for Air Quality Monitoring using Deep Learning and Monte Carlo-based Uncertainty Quantification
No full textIncreasing levels of harmful air pollutants, particularly particulate matter smaller than 2.5 μm (PM2.5), pose a significant global public health risk. While accurate monitoring is essential, a critical challenge persists: high-cost, government-grade sensors are sparsely deployed, while more accessible low-cost sensors lack accuracy. Deep learning models, such as Long Short-Term Memory (LSTM) networks, offer a powerful solution for calibrating these sensors, yet they often lack reliability as they do not report the confidence of their predictions. This absence of Uncertainty Quantification (UQ) undermines trust and can lead to flawed decision-making in environmental management. This study aims to address this gap by separating and quantifying two forms of uncertainty: aleatoric (inherent data noise) and epistemic (the model's own confidence). To achieve this, an LSTM model was developed to predict both a mean value and its corresponding variance by training with a Negative Log-Likelihood (NLL) loss function, allowing it to learn aleatoric uncertainty directly. To estimate the model's epistemic uncertainty, Monte Carlo Dropout (MCD) was applied during inference. By performing multiple forward passes with Dropout active, a prediction distribution was generated whose variance serves as a strong measure of model confidence. An experiment was conducted to find the optimal dropout probability by training and evaluating models with rates from 0.05 to 0.5. While the model with a 0.05 dropout rate achieved the lowest Test RMSE (0.9174) and the 0.35 rate yielded the lowest ECE (0.0228), the results revealed that the model with a dropout rate of 0.10 demonstrated the best overall performance. It achieved the lowest Test NLL of 0.2825, a Test RMSE of 0.9200 (R² of 0.9846), and well-calibrated uncertainty with 97.63% coverage and an ECE of 0.0338. This work presents a validated framework for enhancing the trustworthiness of deep learning models in air quality applications, ensuring that predictions are not only accurate but also reliable
