STEM Student Presentations
April 26th, 2023
12:00 P.M. - 5:00 P.M.
Linda K. Paresky Conference Center
Lightning Talks : Slot #1, 1:00 - 1:30 pm
Lightning Talks : Slot #2, 2:00 - 2:30 pm
Abstracts in Order of Presentation
April 26th Lightning Talks : Slot #1, 1:00 - 1:30 pm
Analysis of Escherichia coli in the Muddy River of Boston, Massachusetts
The Muddy River, located along the Emerald Necklace of Boston, Massachusetts, is a protected, public waterway linked by a series of streams and ponds. Due to sewage cross connections and bird waste the river is subjected to high levels of bacterial contamination, specifically Escherichia coli. The presence of E.coli and other coliform bacteria are strong indicators of poor water quality and have the potential to outcompete bacteria that provide beneficial ecosystem services. This project focused on quantifying e.coli and other coliform bacteria throughout various sections of the Muddy River: the Ipswich Bridge, Willow Pond, Ward's Pond, Leverett Pond, and the bridge at Avenue Louis Pasteur. Water samples were collected, filtered using membrane filtration units, and incubated on MI agar plates for 24 hours before being analyzed under ambient and longwave UV light for presence of bacterial colonies per 100 ml of water. Bacterial growth was most abundant on plates containing samples from Willow Pond and under the Ipswich Bridge. Growth was minimal or absent in the remaining samples, though the presence of fungus was noted on some plates. Areas that yielded high bacterial growth are characterized by poor water quality, though from our experiments it is impossible to tell the exact source of the E.coli. Further studies may indicate the source(s) of the bacteria.
A Qualitative Analysis of Parenting Young Children during the COVID-19 Pandemic
Caitlin Curry & Charlotte Rice
The COVID-19 pandemic challenged many people worldwide, leading to increased stress, anxiety, and depression. The goal of the current study was to understand the unique experiences of parents of young children. 116 parents of preschoolers participated in a larger longitudinal observational questionnaire study following an initial online survey in March 2020 (T1) and at two follow-up periods approximately one and two years after T1. In March-April 2021, (T2), participants were asked to complete three questions: (1) What have you found particularly emotionally difficult about parenting during the pandemic? Please give a specific example. (2) How did you cope with these difficult emotions? Please give a specific example. (3) How, if at all, did these emotions impact your parenting? Please describe using a specific example. Responses to the questions were analyzed using reflexive thematic analysis, involving six steps: (1) Becoming familiar with the dataset, (2) Coding, (3) Generating initial themes, (4) Developing and reviewing themes, (5) Refining, defining, and naming themes, and (6) writing up. We identified six main themes: (1) Overwhelmed and frustrated, (2) Grief over loss of childhood experiences, (3) Challenges of communicating with a young child, (4) Balance between needing to maintain connections and desire to be alone, (5) Pandemic perspective, and (6) Changing parenting style. Findings suggest that interventions focusing on making up for lost developmentally or culturally significant milestones, interventions focusing on mindfulness and acceptance, and skill-building around how to communicate with young children while limiting fear, may all be helpful to parents of young children.
Improving Cognition in Alzheimer's Disease: Evaluating the Efficacy of Levetiracetam (LEV), the Mediterranean Diet, and Physical and Cognitive Training Interventions
Alzheimer's Disease, the most common form of dementia, is distinguished by the progressive impairment of cognitive functioning and episodic memory. It is becoming a rapidly increasing global health concern for the aging population due to the complex nature of clinical presentation and symptom management. Its priority status has led to many areas of research that delve into not only the treatment of the disease but also the importance of disease prevention.
There are a variety of interventions that appear to improve symptoms of cognitive decline and thus the quality of life in Alzheimer's patients. Therefore, it is crucial to examine existing research and recommend further research into the most viable intervention. For example, Levetiracetam, an FDA-approved anti-seizure medication, is currently being investigated through clinical trials to determine if it can prevent cognitive decline in Alzheimer's patients. Physical and cognitive interventions, such as aerobic exercise and paper-and-pencil exercises, have also demonstrated positive effects on cognition in Alzheimer's patients. Additionally, the Mediterranean diet is hypothesized to improve executive and memory functions and slow cognitive decline in those with dementia.
The goals of this thesis are as follows: to describe existing interventions for slowing cognitive decline in people suffering from Alzheimer's Disease, to review published literature about the efficacy of these interventions, and to provide a recommendation for further research into the most viable intervention that could aid in symptom management and slowing disease progression.
Explore Spotify using Shiny: Discover and Analyze Your Favorite Music.
In this project, I developed a Shiny app to help users discover new music and explore their favorite genres with ease. Shiny is an R package that facilitates the creation of interactive web applications from R code. The app uses Shiny to enable users to filter and explore a dataset of Spotify songs according to their preferences for loudness, danceability, genre, track popularity, and artist. The app has two tabs: the first tab displays a table and a list of filtered songs based on user input, while the second tab displays statistical analyses of the dataset. The application features sliders and select input fields that allow users to interact with it. These dynamic tools update the displayed song list and table output in real-time based on the user's inputs. Additionally, the app's interface incorporates custom styles that make it visually appealing and enhance the user experience.
External Ventricular Drain Guide
Hydrocephalus is a condition in which the brain's ventricles swell from a buildup of cerebrospinal fluid due to head injury or problems with prenatal development. The procedure to treat hydrocephalus is called ventriculostomy, where a hole is drilled in the skull and an external ventricular drain (EVD) is inserted to help relieve the intracranial pressure that is built from the excess cerebrospinal fluid. This treatment relies on a freehand insertion of the drain, which leads to multiple insertions and potential harm to the brain.
The aim for the Extraventricular Drain Guidance project is to develop an accurate method to improve current ventriculostomy procedures by using ultrasound to detect ventricles and create an improved guide to aid neurosurgeons and physicians with hydrocephalus. Through the use of a model brain (made of PVA and graphite) and transducers between 2.25 MHz and 10 MHz, we conclude that the ventricle can be accurately seen through using ultrasound.
Exploring the electrochemical reduction of CO₂ using colloidal thiospinel nanoparticles
For this project, I have synthesized nanoparticles of the FeCo2S4 thiospinel using colloidal synthetic techniques. I characterize the nanoparticles with different technologies such as X-ray fluorescent (XRF) spectroscopy, dynamic light scattering (DLS), scanning electron microscope (SEM) imaging and X-ray diffraction (XRD). In order to get an accurate measurement of each element in a sample of the thiospinel I synthesized, I had to create a calibration curve for the XRF. DLS was used to determine the range of sizes of the particles. XRD was used to characterize the crystalline structure of FeCo2S4.
Lightning Talks : Slot #2, 2:00 - 2:30 pm
Design and construction of a synthetic mycobacterium phage genome using transformation-associated recombination (TAR) cloning in Saccharomyces cerevisiae.
The rise of antibiotic resistance in pathogenic bacteria is driving a renewed interest in bacteriophages. These naturally occurring antimicrobial agents can potentially be exploited as novel therapeutic agents. Existing limitations such as bacterial resistance and narrow host range; however, are problematic. The advent of new techniques in genome science and synthetic biology have provided mechanisms to build unique phage genomes that confer new functions that may overcome these difficulties.
Building upon methods developed in the Build-a-Genome course at Johns Hopkins University and the Synthetic Yeast Project (www.syntheticyeast.org), we present the construction of a synthetic genome of the mycobacterium phage Giles. We adapted the novel assembly line approach introduced by Oldfield et al in 2017 and set out to construct the complete genome from 12 overlapping DNA fragments of the phage using the yeast based transformation-associated recombination (TAR) system. Individual mini-chunks of DNA are incorporated via polymerase chain reaction (PCR) with a TAR cloning vector using specifically designed primers. The DNA fragments are then assembled into the full length phage genome inside yeast. Genetic modifications can be made by attempting to insert novel DNA sequences into one or more fragments.
The final objective of this project is to transform the synthetic phage into bacterial cells and measure its infectivity. The results of this project will show a successfully engineered and infectious clone of phage Giles. The significance of which will be to demonstrate the optimization of novel methods to construct a synthetic phage genome.
Machine Learning in Prosody and Meaning
Madeleine Guettler, Sofia Hischmann, Shirley Fong
Prosody in spoken language is widely thought to convey meaning, but the complexity and variety of prosodic renditions have made modeling a specific prosody-meaning mapping elusive. In this work, 238 prosodically labeled utterances are used to classify speech as either exclamative or neutral. A first pass was made to cluster accent types (emphasized words) into three groups. This research project captures the importance of a variable number of loosely categorizable elements; clustering is performed; and the percentage of accent type is used as an attribute in the final random forest classification step.
The Impacts of Inspiration Porn
My study explores the impacts of inspiration porn. Inspiration porn is a representation of disabled people characterized by images of visibly disabled people performing physical activities similar to those of able-bodied people with minimal accommodations accompanied by short motivational captions that are meant to encourage viewers (Hadley, 2016). Although it is important to have disability represented in popular culture, inspiration porn may have unanticipated negative consequences. Since inspiration porn depicts disabled people achieving athletic goals with minimal assistance, it may undermine understanding of and support for necessary disability accommodations. Based on theoretical work surrounding similar topics (such as Jost & Hunyday, 2005), I expected that inspiration porn would increase system justification, and decrease support for disability-related social change.
296 participants were recruited to take an online survey and randomly assigned to see either an example of inspiration porn or a control photo. Participants completed a series of measures, including measures of system justification, dehumanization, and support for disability-related social change. No significant effects of exposure to inspiration porn on system justification or social change were obtained, but there was a significant effect on dehumanization. Participants in the inspiration porn condition viewed disabled people as less human than participants in the control condition. This study represents the first known quantitative exploration into inspiration porn and its potential negative impacts on disabled people, such as increased dehumanization. It highlights the need for further research on the topic of inspiration porn as well as the need to allow disabled voices to guide psychological research.
A Framework to Develop Automatic Speech Recognition for Low Resource Languages
Nardos Alemu, Chelsea Hua
Current Automatic Speech Recognition (ASR) systems, like Google Assistant, Apple's Siri, or Amazon's Alexa, continue to only support a small number of languages (English, Mandarin, Arabic, etc.), primarily those spoken in developed nations with abundant resources. While these languages have been able to reap the benefits of having such technology at their disposal, places like Ethiopia, and Vietnam are still far behind. This work represents a global collaboration to create a framework for customizing ASR systems for low resource languages (LRLs), or languages with limited human and financial resources. This paper describes the methodology for using an existing application (Kaldi) to implement an ASR system for two such languages, Amharic and Vietnamese, with the least amount of annotated speech. The languages are chosen to leverage available student expertise and create cross-cultural connections. The objective of the research is to create a procedure by which, given enough training records and annotation, any language can be added.
The effects of relaxing and energizing music and personality traits on mood states
Sarah Kim, Grace Lincoln
The emotional effects of music listening have been previously studied with an eye towards understanding how music can be employed to enhance emotion regulation. The present study will examine how individual differences in personality influence how people select and emotionally-respond to music. As part of a larger study, college student participants were randomly assigned to select a song they find energizing (condition A) or relaxing (condition B) during a single laboratory session. Participants completed a baseline measure of self-reported personality traits and mood. Participants listened to their selected song and then completed a post-task assessment of mood and a post-task questionnaire including demographic information.
We hypothesized that energetic music will result in greater increases in positive affect than relaxing music and this effect will be strongest among individuals scoring high in extroversion. We further hypothesized that extroversion will be associated with a tendency to choose songs with emotionally-positive lyrics which will in turn predict greater reductions in depressed mood following music listening. In contrast, neuroticism will be associated with a tendency to choose music with more emotionally-negative lyrics which will predict less reduction in depressed mood following music listening. Data collection is ongoing and results will be ready to present in time for the STEM Undergraduate Presentation Event. Results will be discussed in terms of potential applications of music to support psychological well-being.
The Effects of Economic Status on Academic Performance
Audrey Parker, Nicki Nicholas, Becca Walz
Our project aims to predict how a student will perform on standardized tests given indirect factors associated with low income. The dataset “Students Performance in Exams” from Kaggle gives three indirect factors: students' parental academic level, whether or not a student had free/reduced lunch, and whether or not a student completed test preparation. There is a strong correlation between these three factors and level of income. Research has shown parental academic level affects children. As the level of education increases, the ability for parents to be with their children, model achievement-oriented behavior, and provide age-appropriate engaging activities also increases (Penn State). The second indirect factor of low income used in this project is poor nutrition. As a way to ensure students receive nutritious lunches, schools provide lunch and offer reduced price or free lunch for families who can't afford it. However, outside of school, low income households can struggle to get regular nutritious meals. This lapse has profound effects on students' cognitive performance (Brooks-Gunn). Lastly, test scores are directly affected by the student's preparation and research supports the idea that income directly affects how well a student can prepare for the SAT (Kantrowitz, M.).
After cleaning, organizing, and analyzing the data the relationships were shown to be consistent with the research. Thus, we continued on to create our algorithm. The algorithm first splits the data into a training set and a testing set. The training set is used to teach the classifier. Our classifier was a Gaussian Naive Bayes. This classifier was chosen because it assumes the independence of each variable and allows us to add together the separate probabilities. Then, the testing set was used to test the accuracy of our classifier. The algorithm was about 14% more accurate than a random guess. As with most machine learning algorithms, this algorithm would improve through incorporating additional data or using larger training sets. Additionally, integrating another classifier would improve the accuracy. Algorithms that aim to understand and combat economic disparities are exceedingly important today because of the limited educational support that currently exists for students from a variety of backgrounds. In order to be able to stop the pattern of inequity for students of lower income we need to continue to conduct research that investigates the causes.