Gemstone Team Research

Permanent URI for this collectionhttp://hdl.handle.net/1903/9070

The Gemstone Program at the University of Maryland is a unique multidisciplinary four-year research program for selected undergraduate honors students of all majors. Under guidance of faculty mentors and Gemstone staff, teams of students design, direct and conduct significant research, often but not exclusively exploring the interdependence of science and technology with society. Gemstone students are members of a living-learning community comprised of fellow students, faculty and staff who work together to enrich the undergraduate experience. This community challenges and supports the students in the development of their research, teamwork, communication and leadership skills. In the fourth year, each team of students presents its research in the form of a thesis to experts, and the students complete the program with a citation and a tangible sense of accomplishment.

Browse

Search Results

Now showing 1 - 10 of 203
  • Thumbnail Image
    Item
    Affordable Real-Time CV for Disaster Relief and Beyond
    (2025-05) Linsider, Eitan; Rana, Jay; Roy, Nathan; Stern, Jacob; Deane, Anil E.
    Search and rescue operations following natural disasters are critical for saving the lives of individuals trapped and in need of aid. However, many of the technologies used in search and rescue (helicopters, planes, boats, etc.) can be prohibitively expensive for nations with low GDP. This issue is only exacerbated by the higher number of deaths due to natural disasters in such nations due to less resilient infrastructure. Unmanned Aerial Vehicles (UAVs) have the potential to replace more expensive technologies in the search for endangered individuals and analysis of at-risk hardware without endangering rescue personnel. While advancements in low-cost UAVs for commercial and hobby use in addition to the development of lightweight computer vision (CV) software and dedicated processors have set the stage for low-cost search and rescue UAVs, stand-alone UAV costs and part scarcity still pose challenges. Team ANDRR developed a real-time CV module for UAVs to aid in natural disaster relief using a wide range of compatible commercially available parts wherever possible. By conglomerating existing systems into a single effective and easy-to-use unit, we were able to identify individuals in real-time at low cost. The development guide and software for the module are provided to provide increased access to real-time CV processing for search and rescue and other operations.
  • Thumbnail Image
    Item
    Fast Prediction of Full Quantum Dynamics with Deep Recurrent Neural Networks
    (2025-05) Anand, Yash; Gouzoulis, Mark; Kingston, Linfred; Kugelman, Jander; Li, Jonathan; Pham, Dustin; Siddalingaiah, Eashan; Taylor, William; Witt, George; Dutt, Avik; Hafezi, Mohammad
    Numerical simulations of interacting quantum systems are computationally very intensive, typically requiring resources that scale exponentially in the number of particles. A plausible approach to overcoming this unfavorable simulation time is to train deep neural networks over short timescales and use them to infer dynamics over much longer timescales. We demonstrate that such a speedup is possible using deep recurrent neural networks, including LSTM and Transformer-based networks, by predicting the quantum dynamics of multiple ’classic’ systems - the Ising, Heisenberg, and Hubbard models, with up to 9 spins, at most. We observe up to 3 orders of magnitude of data generation speedup for systems that can still be simulated with full system evolution. We observe this same performance - O(0.1) seconds - to generate data samples that cannot be generated with the resources we have available to use. Unique to our work, we predict the full wavefunction dynamics of the systems, which can then be used to calculate the evolution of measurable and theoretical observables over time. We present sample predictions for our models and compare the efficacy of the different approaches with varying context-length for prediction, spin count, and Hamiltonian parameters (mixing, interaction strength, etc.), at best accurately predicting (< 10−6 mean square error - MSE) up to 90% of a single period with 10% of a period for context. We probe a number of frustrations, including square and triangular interaction lattices, more complex next-nearest neighbor interactions, etc. to understand what currently limits strong machine learning - ML - results in this space. Our results indicate that the primary inhibitor to fast prediction at scale is the system scale, not the complexity of the dynamics. We anticipate that our work will provide insights towards extending the coherence time of quantum systems such as qubits and spins by determining the issues that stand in the way of network training and prediction on realistic Hamiltonians. We further believe that this work has immediate application in the simulation of large-scale neutral atom arrays, like Yt-171, under so-called Lieb-Robinson bounds.
  • Thumbnail Image
    Item
    The Relationship Between Gut Motility and the Phases of the Menstrual Cycle
    (2025-05) Anders, Tara; Behailu, Eliya; Callaway, Delaina; Gardiner, Kara; Grant, MAggie; Qian, Christina; Redmond, Leah; Sany, Nakati; Sarkar, Grace; Ullah, Tasnim; Hall, Brantley
    Female reproductive hormones fluctuate significantly throughout the menstrual cycle, influencing gastrointestinal motility and function. These changes in gut motility contribute to gastrointestinal symptoms such as flatulence, diarrhea, and constipation. When transit time is faster, fewer carbohydrates are absorbed in the small intestine, leading to increased microbial fermentation and gas production in the large intestine. Given the established effects of reproductive hormones on gut motility, we hypothesized that we would detect increased gut microbial gas production during menstruation, when estradiol and progesterone are lowest and transit time is fastest. However, due to a lack of tools to measure gut microbial gas production longitudinally, the literature lacks a formal test of this hypothesis. Therefore, we used a novel tool, the Smart Underwear device, to measure gut microbial gas production as a proxy for intestinal transit time. Participants wore the device for a total of nine days over the course of one menstrual cycle. On device-wearing days, participants logged the meals they consumed. Additionally, participants used Luteinizing Hormone test strips in the middle of their cycle, approximately on day 14, to confirm ovulation timing and more precisely map menstrual cycle phases. We extracted and analyzed the data from our returned Smart Underwear devices to measure gut microbial gas production as an indicator of intestinal transit time at different phases of the menstrual cycle, providing the first longitudinal measurements of gut microbial gas production across the menstrual cycle. Initial findings suggest that microbiome activity varies significantly across phases, with heightened activity during ovulation. Our analysis revealed that higher concentrations of flatulence peaks occurred during the ovulation phase.
  • Thumbnail Image
    Item
    Examining Contributors to Black Maternal Health Experiences in Prince George's County, Maryland
    (2025-05) Eskinder, Aden T.; Nnabue, Alexis C.; Onyekwere, Christal C.; Battish, Isabella H.; Owie, Esohe T.; Harris, Madison L.; Redwood, Sidney A.; Fishkin, Stephanie M.; La Touche-Howard, Sylvette
    Black women in the United States face a maternal mortality rate three times that of white women, a disparity mirrored in Prince George’s County, Maryland (Hoyert, 2023; Maryland Department of Health, 2022). In 2019, the Prince George’s County Health Department reported that between 2008 and 2017, Black, non-Hispanic mothers experienced the highest pregnancy-related maternal mortality rate (37.4 deaths per 100,000 live births) in the county (“Maternal Infant and Health Report,” 2019). This study explores how maternal access to healthcare and provider cultural competency training influence maternal health disparities in Prince George’s County. Using a mixed-methods approach, the research team surveyed and interviewed two key populations: Black mothers ages 18-34, residing in Prince George’s County, Maryland, and maternal health providers practicing in the District of Columbia, Maryland, and Virginia (DMV) region. By examining patient-provider interactions and gaps in medical education, this research aims to inform policy and curricular reforms to improve Black maternal health outcomes.
  • Thumbnail Image
    Item
    Hearing Aid Audio Processing Model Benchmarking with Binary Environmental Classification
    (2025-05) Mukhopadhyay, Rajit; Tumkur, Bhargav; Li, Lily; Waters, Samuel; Reyes, Chelsea; Sarkar, Ronoy; Nair, Rahul; Patel, Pruthav; Sare, Perfect; Shah, Sahil
    Current hearing aid audio processing models are trained to filter out noise and amplify specific sounds such as speech. However, real-world audio environments contain a multitude of sounds that users may want to hear that cannot exhaustively be defined within the model. Team ECHO’s project proposes a method for hearing aid audio processing models to account for these unknown sounds without explicit definition. The team developed an open source audio dataset containing unfiltered-environmental audio that is categorized by a set of 4 binary features (indoors, crowded, walking, speaking) which are applicable to any audio environment. The team collected audio data using over-the-ear microphones at different locations centered around the UMD College Park campus and in Washington DC. The team also constructed a benchmarking platform for researchers to compare the performance and efficiency of different models in different environments based around this dataset. The platform evaluates models on 6 different metrics (e.g., Noise Floor, Signal-to-Noise Ratio, Dynamic Range, and Crest Factor). Results from testing various audio processing models demonstrated significant differences in performance metrics across unique noise environments. The developed and tested benchmark serves as a groundwork for future audio processing research tailored specifically for hearing aid comfort and realism.
  • Thumbnail Image
    Item
    Firefighting Exoskeleton to Reduce Stress and Strain
    (2025-05) Bigot, Tom; Bosco, Connor; Ingram, Brett; Mense, Jessica; Salanitiri, Nicholas; Smith, Liam; Spriggs, Donald; Sunderland, Peter B.
    Overexertion stress and strains and musculoskeletal disorders are the leading causes of injury in firefighters. Lower back injuries resulting from lifting are prevalent in both fire and Emergency Medical Service (EMS) calls. Exoskeletons which provide specialized support during repetitive movements have been increasingly implemented in construction, manufacturing, and healthcare settings in recent years. However, the unique temperature, weight, fit, and quick donning requirements of an exoskeleton suited for use in firefighting renders current commercial exoskeletons unsuitable for firehouse implementation. Team Exo aimed to close this gap in research by exploring the requirements and testing methods of a firefighting exoskeleton. Several iterations of the exoskeleton were completed and categorized into three distinct prototypes. The exoskeleton consisted of dual-spring housings, leg attachments, and an upper body harness. The spring system worked with a cable and plunger to compress the spring while squatting, and decompress when rising; resulting in a loading of the spring on descent, and an assistive lifting force on ascent. Several testing phases were completed, including competing at the American Society for Testing and Material (ASTM)’s 2025 Exo Games and further testing at the National Institute of Technology (NIST). The final prototype of the exoskeleton functioned as a facilitator for the gluteus and hamstring muscles that assisted the user in lifting. Testing focused on user-reported comfort and perceived assistance through a wide variety of firefighter specific movements and followed the current industry-wide standards of exoskeleton testing. Further testing with electromyography (EMG)s, models, and cardiopulmonary exercise testing (CPET) would be beneficial, and should be expanded on when more comprehensive testing standards are established. Further modifications for compatibility with firefighting gear and material improvements would also be necessary for firehouse implementation.
  • Thumbnail Image
    Item
    Investigation of Fungal Extracts as Anti-Biofilm Agents
    (2025-05) Bao, Lili; Cardillo, Andrew; Chowdhury, Ranita; Crowley, Leah; Eldo, Annette; Gibney, Catalina; Gul, Beyza; Kong, Matthew; Meda, Anju; Meyers, Sumangal; Poulin, Myles
    Many nosocomial infections are derived from microbial growth on implantable medical devices that form biofilms. Biofilms are complex colonies of bacteria that can grow on the surfaces of medically implantable devices and typically have antibiotic-resistant properties. Currently, no commercially available FDA-approved or EPA-certified biofilm sterilants are biocompatible for medically implantable devices. This study examined potential applications of shiitake (Lentinula edodes) and turkey tail (Trametes versicolor) water-soluble fungal extracts as preventative antibiofilm agents. Through the isolation of extracts derived from turkey tail and shiitake mushrooms, this research sought to find a sustainable, safe, and accessible method to treat biofilm infections originating from medically implanted devices to address the rise in antibiotic resistance, in addition to other practical applications. Our results show that ultrasonic-assisted aqueous shiitake extract inhibits Staphylococcus epidermidis biofilm formation in a dose-dependent manner and displays general cytotoxicity, while ultrasonic-assisted aqueous turkey tail extract shows no significant biofilm or growth inhibition. Further research is needed to characterize these isolates and determine their particular mechanisms of action.
  • Thumbnail Image
    Item
    Cleansing Data and Bias within Predictive Policing Algorithms
    (2025-05) Arellano, Trina; Chen, Alex; Du, Allen; Eichstadt, Andrea; Lin, Aaron; Samuels, Nicole; Tao, Grace; Tasneem, Zoya; Versace, Rios; Taghi Hajiaghayi, Mohammad
    Hot spots policing—the allocation of police resources toward high-crime areas—has been revolutionized by machine learning. Instead of relying on historical crime hot spots, predictive policing algorithms allow departments to allocate officers to where crime is expected to occur next. This has led to their increasing adoption by especially large police departments, as well as modest reductions in crime. However, predictive policing algorithms have thus been shown to exhibit similar biases to traditional policing methods. A vast literature has shown that nonwhite areas are more frequently policed, and that laws are disproportionately enforced against nonwhites in these communities. This creates a problem for predictive policing; since these algorithms are trained on historical crime data which reflects these racial biases, predictions come to perpetuate racial bias into the future. As such, our team has built a new predictive algorithm which not only uses more contemporary machine learning techniques, but directly accounts for demographic fairness in its predictive judgments. Using real crime data, we then tested our model against PredPol, a state-of-the-art predictive policing software, comparing them on predictive accuracy and on racial bias in their predictions. Our results showed that our model outperformed PredPol in both predictive accuracy and fairness, demonstrating that it is possible to make policing more equitable without sacrificing the predictive accuracy of these algorithms.
  • Thumbnail Image
    Item
    Assessing the Biomechanical Properties of ErythroMer for Partical Integrity in an ex vivo Organ Perfusion System
    (2025-05) Bannerman, Joel; Kettula, Claire; Lang, Jackson; Passaro, Emily; Simcox, Talya; Somerville, Matthew; Doctor, Allan
    The development of hemoglobin-based oxygen carriers (HBOCs) is critical for advancing organ perfusion systems. This study evaluates the structural integrity of ErythroMer (EM), a novel oxygen carrying nanoparticle, as a potential blood substitute for ex vivo kidney perfusion systems. Shear stress assessments were conducted to simulate physiological forces within complex pump systems. This study was organized through three scientific aims: quantification of EM peak tolerance, assessing shear stability over time, and evaluating the impact of suspension medium on shear tolerance. These aims assessed EM shear stability as a safety metric, ensuring maintenance of particle integrity were it to be used as a perfusion solution within an organ pump system. The results demonstrated that EM maintains structural integrity and resistance to lysis, even under high levels of shear stress. These findings support the potential use of the EM particle in future organ perfusion pump systems and its broader applications in biomedical fields.
  • Thumbnail Image
    Item
    Investigating combination of photodynamic therapy and AXL inhibition for improved treatment outcomes of glioblastoma
    (2025-05) Fadul, Nada; Mitchelmore, Louise; Farrag, Farah; Shaw, Anna; Gangar, Dilan; Yeon, Jennifer; Pang, Sumiao; Hays, Rebecca; Huang, Huang-Chiao
    Glioblastoma multiforme (GBM) is notorious for its aggressive behavior, brain invasion, and poor prognosis, with a median survival of less than 18 months. Limited treatment options contribute to GBM’s status as one of the deadliest cancers. The intricate tumor composition and invasion of vital brain areas render it resistant to standard therapies, necessitating novel approaches to extend post-diagnosis survival. Intraoperative photodynamic therapy (PDT) has emerged as a promising technique, involving the administration of a photosensitizer before surgery and red light application afterward to target residual tumor cells. Recently, an excipient-free nanoparticle formulation of verteporfin (NanoVP) photosensitizer was developed for PDT of GBM, demonstrating superior efficacy in reducing tumor burden and extending animal survival compared to existing photosensitizers. We explored the combined effects of NanoVP-PDT and clinically promising AXL inhibitors on GBM cells. Phospho-AXL, which is highly expressed in GBM tumors and correlates with shorter overall patient survival, represents a compelling therapeutic target for small-molecule inhibition. In this study, we investigated the anti-GBM effects of combining NanoVP-PDT with AXL inhibitors in vitro as a new treatment approach to combat GBM.