| Abstract # 2 |
| Boykin, T; Resto, L; Mayer, A, Veteran isolation and perceived mental distress in the South Carolina Lowcountry, Nursing and Health Professions |
| Background: This research aims to explore how social and geographic isolation contribute to mental distress among military veterans reintegrating into the civilian community in the Low country. According to recent data, approximately 16% of military veterans report mental health issues. Veterans often link experiences in war such as killing and injuring assailants and witnessing the death or injury of other service members to feelings of mental distress. In addition, veterans report increased mental distress when transitioning back to civilian life after military service. This project aims to provide preliminary evidence supporting the need for improved integration of veterans into the general population. Methods: We utilized a qualitative study design employing semi-structured interviews based on a grounded theory approach. Between January and March 2020, 20 interviews were conducted face-to-face with military veterans within five years of leaving active military duty. An iterative thematic analysis of the data identified areas of consensus and divergence. Grounded theory guided the coding of each transcript, with analysis informing the next round of interviews and the final codebook. Results: The preliminary results indicate that mental, emotional, and financial challenges among military veterans transitioning from active military service to civilian life are substantial. Feelings of isolation and difficulty socializing with the broader community are potential factors for the development of mental distress in this population. On the other hand, Veteran Resource Centers (VCRs), particularly those centers housed within educational institutions, appear to mediate feelings of social isolation and mental distress among those who can access them. Conclusions: Our research demonstrates the potential benefit to new veterans of local-level Veteran Resource Centers (VRCs), especially those housed in educational institutions of the Low country. The support of other military veterans, regardless of branch, appears to be a mediating factor in perceived levels of mental distress within this population. |
| Abstract # 3 |
| Nicholas Ferry, Coronavirus Topic Trend in Twitter, Computer Science |
| COVID-19 has been and is continuing to have a significant impact on the American way of life. Twitter is a social media platform that allows users to express their opinions concisely due to the 140 character-limit per post. These individual opinions expressed through the Twitter platform are known as Tweets. This research attempts to analyze Tweets to provide an understanding of how the population is reacting socially to this pandemic without any filters on their opinion. The most influential Tweets can be collected by identifying the most impactful epochs of this pandemic. In addition to the pandemic epochs, significant media reporting events were used to analyze the reaction of the American population to these media Tweets. The media companies used were: NBC, Fox News, ABC News, CBS News, CNN, MSNBC, New York Times, and VICE News. Sentiment analysis is classifying the subjectivity rating of a sentence into three possible categories: positive, negative, and neutral. The subjectivity ratings are based on the polarity of the sentence that is firmed by the individual subjectivity of each word within the sentence. Essentially, the more positive words in a sentence, the higher the polarity thus resulting in a positive subjectivity. This is true for the reciprocal of negative subjectivity as well as neutral subjectivity. By applying sentiment analysis to the individual Tweets retrieved, an understanding can be established of the subjectivity rating of the general opinion of the American population. This research analyzes the sentiment during pandemic epochs as well as over a broad time to understand the changing opinion of the American population. |
| Abstract # 4 |
| Lily Coutu, Brenna Dickerson, Aliyah Dunn, Sarah Elliott, Audrey Rucker, Davonte Saulsberry, Swati DebRoy, PhD., Spatial Analysis of Association and Impact of Childhood Obesity Rates on Environmental Factors in Beaufort County Elementary School District, Mathematics |
| Obesity is an increasing health issue for children in the United States and South Carolina is ranked 7th among the 50 states for obesity. Being obese or overweight is a risk factor for a lot of health issues including diabetes, heart problems, hypertension and for a lot of individuals it is detrimental. Factors related to genetic makeup have been highly correlated to obesity, but an appropriate amount of physical activity along with diet modification is known to improve overall health outcomes, especially in children. Although Beaufort County is declared as the ‘healthiest County’ in South Carolina, the prevalence of overweight and obese children can vary greatly among schools in this county. This study attempts to parse out links between the prevalence of obesity in 15 schools in Beaufort County School District and built-environments in the zone it serves. Data on factors like frequency of grocery stores and recreational facilities were collected from Google Maps, designation as a food desert from S.C. Department of Health and Environmental Control (DHEC). These factors along with race, gender, socio-economic status are used to predict the prevalence of obesity through multiple regression analyses. We also examine the impact of the same factors and prevalence of obesity on the academic performance of the school determined by SC Ready Math and SC Ready English Language Arts test scores. |
| Abstract # 5 |
| Nicholas Ferry, James Cheatham, Richard Herrin, Parallelizing the Support Vector Regression Algorithm, Computer Science |
| A prediction algorithm analyzes data input and creates an output or predictions based on the input data. Linear regression is a prediction algorithm that uses the best fitting line of the data model. Support vector regression is a variation and improvement over linear regression that has the capability of adding limits of the acceptable error to either side of the linear regression line fit. In this project, we first implement the serial version of the support vector regression algorithm capable of running on a single core. To test the prediction algorithm, we collect mortality rates of various significant viruses and use them to predict the mortality rate of the COVID-19 pandemic. We then parallelize the support vector regression prediction algorithm so that it can run on multiple processing cores. The parallel execution of the prediction algorithm on multiple cores will help improve the performance compared to the serial version of the prediction algorithm. |
| Abstract # 6 |
| Alexis Miller, Katherine Redmond, Bradley Lamb, Serial Vs. Parallel: A comparison of algorithm performance, Computer Science |
| Understanding the performance differences between serial and parallel algorithms can help to determine which version of the algorithm is appropriate for each situation. Our goal is to learn the differences between the performance of serial and parallel algorithms and to determine if each variant has an appropriate situation to be used. We hope to achieve this by studying each of our serial algorithms carefully and theorizing their parallel conversions with Ian Foster’s method. We will analyze each algorithm in its most basic form to understand how it works and why it operates at its current efficiency, then decide how we would convert it to a parallel version. We’ll analyze how the parallel version would differ and how we believe the efficiency would change. Then, we’re determining when, if ever, the conversion makes sense to use. We hope that by understanding the algorithm in both forms and the conversion method itself, we can determine if parallelization makes sense for that algorithm and how to accomplish it. |
| Abstract # 7 |
| Veronica McLeod, Frank Cazales, Lingtao Chen, Shae Gantt, Brice Adkins, Rasheed Dias, Makayla Keasie-Woo, Modeling the Spread of Covid-19 in the State of South Carolina and the High-Incidence Counties, Mathematics |
| SARS-CoV-19 is a type of coronavirus that causes an infectious respiratory disease called Covid-19. It was first identified in Wuhan, China in December 2019 and quickly spread throughout the globe. On January 21st, 2020 the United States identified the first confirmed case and South Carolina detected their first two presumptive positive cases on March 6th, 2020 that was later confirmed by testing. The South Carolina Department of Health and Environmental Control (SCDHEC) has tested and reported cases found in the state every day as well as presented guidelines for citizens, healthcare facilities, and governments to follow. South Carolina has declared a State of Emergency in order to implement emergency response procedures in addition to new temporary legislation for “Social Distancing†with the goals to improve detection of infected individuals, assist health care facilities and first responders in better handling cases of infected patients, and intervening to reduce person-to-person contact of individuals predicting this will slow the spread of Covid-19. In this project we apply a basic Kermack-McKendrik ordinary differential equation model to this pandemic in the state of South Carolina and the top high-incidence counties including Beaufort and Charleston counties. Using the known recovery period of COVID-19, the transmission parameter is estimated for the regions through least-square fitting to the daily reported cases. Based on this estimation, we also calculate the projected magnitude of the peak and time needed for the infection to subside if no change in control measures occur. We expect to study the impact of change in control measures (for social distancing) directly targeted to reduce transmission. In the future we intend to expand the model to include death due to infection and an asymptomatic yet infectious class in the population. |
| Abstract # 8 |
| Cynthia Flowers, Corona Virus Case Study: An Investigation of Doubling Rates, Computer Science |
| Corona Virus is spreading in many countries in the spring of 2020. It has been the major crisis all over the world. It is crucial to predict how the number of cases increases over time in a country. Since the virus spread over countries such as China and Italy earlier than the U.S., the increasing models in these countries can be potentially used to predict how the virus increases in the U.S.. However, it is hard to relate the number of cases over different countries, because the difference in population and other factors. In this work, we attempt to analyze the increasing trend based on doubling rate. Doubling Rate is defined as the reciprocal of Doubling Time. In practice Doubling Times are used to measure growth rates: how long does a measured quantity take to double? A larger or longer doubling time means slower growth while a shorter doubling time indicates rapid growth. Doubling rate gives a better visual representation of decreased growth rate for the purpose of an expository graph, as it decreases with decreasing growth How do you “flatten the curve?â€: send the doubling rate hurtling towards zero. |
| Abstract # 9 |
| Breona Davis, Emani Dente, Deanna Waid, Shane McCarty, Fair Market Rent Trends in South Carolina, Mathematics |
| Fair Market Rent Trends in South Carolina Abstract The purpose of this project was to understand how and why the Fair Market Rent for a 2-bedroom apartment changed in South Carolina in the past 10 years. We also wanted to understand whether or not there are any meaningful trends in this data. To understand any potential trends in Fair Market Rent (FMR), we decided to compare the Fair Market Rent Values from twenty randomly selected South Carolina Counties against other data sets such as population, and Median Family Income (MFI). For example, one of the counties that were randomly chosen was Beaufort County. In 2018, the Fair Market Rent was $1,056 for a two bedroom as compared to $887 in 2010. For this county, the FMR increased as well as the MFI and the population in the same years. In 2018 the MFI was $72,200 as compared to $66,400 in 2010. A correlative examination of any potential trends between these data sets, along with a predictive analysis of any trends uncovered may provide valuable insight for residents; helping them better understand their socioeconomic environment and factors which may affect rents in their local area. |
| Abstract # 10 |
| Kevin Balderas, Alex Rendon Jonguitud, Susan Reyes Garcia, Eileen Silva Rayo, Swati Debroy, Deaths by Diarrhea in the America’s, Mathematics |
| In the Americas, the number of people dying because of diarrhea are significant with no major improvement occurring. Diarrhea is a condition in which, aliment is rapidly passed through the gastrointestinal (GI) tract causing malabsorption of necessary nutrients and liquids. Leading to excess dietary excretion. There are many causes that lead to this condition; GI infection from bacteria, parasites and viruses, malnutrition, contaminated water sources and poor hygiene practices. Although diarrhea is often an unforeseen cause of death, according to Our World in Data it was placed number eight globally in 2017, when 1.57 million people died because of this condition. The number of deaths has been decreasing throughout the years but not significantly enough to cause a major difference. Diarrhea is the second leading cause of death in children under the age of 5 with 525,000 children dying every year, according to World Health Organization. The number stood out making us realize that we needed to bring more awareness to the people of the impact of diarrhea. So, our study focuses on the identification of how different political, socio-economic and climatic factors correlate with the death caused by diarrhea. We also want to see if one co-variant has a higher impact than the other. This would all help bring awareness to people on what could be changed to help decrease the number of deaths caused by diarrhea. |
| Abstract # 11 |
| Zachary Shafer, Examining Data Mining Classification Methods Through Weka, Computer Science |
| Data mining allows us to gather new information from a dataset. While there are several methods of gathering new information through data mining, this project examines the use of classification methods. The dataset used in this project consists of 1,025,010 instances and eleven attributes from poker playing cards. The first ten attributes in sets of two represent the suit and rank of five cards respectively. The final attribute represents the “poker hand†that would be formed by the five cards. This dataset is examined using Weka, an open-source, data mining software, and the results of different classifications methods within the software are observed to determine which method is most efficient and most accurate. |
| Abstract # 12 |
| Chandler Chapman, Tania Pegues, Lincoln Fuller, Donna Daniels, McKayla Cavanagh, Marino Sorbara, Impact of Population Increase on Main Rivers of Beaufort County, Mathematics |
| Water quality around the world is depleting, and one of the main reasons is from population growth and new development. When looking at livable land area to population, Beaufort county has one of the highest population densities in the US. To see how this is affecting the water quality of the three main rivers in Beaufort county, the Broad, Beaufort and Chechessee rivers. The group looked at many response variables of water quality. According to USGS.gov, excess nitrogen and phosphorus in the rivers and waterways may cause a number of adverse health and ecological effects, including eutrophication and hypoxia. The main source of excess nitrogen and phosphorus in fresh and ocean water are stormwater runoff and human waste from septic tank leakage. Ph was examined as well because as the local rivers become more acidic the crustaceans, that are very popular in the lowcountry, will be the first to die off. Lastly, dissolved oxygen data was examined to show the inverse relationship it has with nitrogen content and how it may bring about hypoxia in our rivers. Low dissolved oxygen levels are ecologically detrimental as it provides a breeding ground for pathogens. As population growth, housing development and commercial fuel use continues to increase in Beaufort county, the land’s natural absorption abilities will decrease and increase major stormwater runoff and septic leakage issues. In this project, the group applied many different linear regression models and other models to determine the impact of population increase on Beaufort county’s main rivers from 2002 to 2016. The goal of this model is to aid the government in determining water quality monitoring locations in Beaufort county. |
| Abstract # 13 |
| Jay Cheatham, Running a Linear Regression Algorithm to Determine a Relationship with the NBA 2017-2018 Regular Season Statistics, Computer Science |
| In a previous course taken with Dr. Brian Canada, I had taken a dataset that had complied the statistics from the National Basketball Association (NBA) 2017-2018 (by gathering the data from Basketball Reference) to see if there was any relationship between two of the statistics by running a linear regression algorithm through a Python script. With the permission of Dr. Xiaomei Zhang, the same set of data is being ran through Weka (an open source data mining application) by converting the original Comma-separated values (CSV) file to a Attribute-Relation File Format (ARFF) file to analysis the same hypothesis from the previous course further by mining the data from the statistics of the NBA season and running a linear regression test for multiple attributes (more than the initial two from the original test). After mining the data and initializing the linear regression algorithm, it’s been determined that there is a relationship between the statistics of each player (though there are a few outliers); because the linear regression algorithm was performed with more than two attributes (unlike the original test from the previous course), it strengthens the conclusion from the previous course that a majority of the attributes are dependent on each and do have a relationship, though there is more evidence of it now and the number of outliers has decreased. |
| Abstract # 14 |
| Kaitlyn O`Hearn, Perception of Piercings, Social Sciences |
| This research examines how individuals perceive people who have facial piercings. Individuals filled out information regarding whether they had any facial piercings along with a shortened version of the Big Five personality survey. Then participants filled out a questionnaire for two photos. In the first photo, the woman had no facial piercings and in the second photo, the woman had three piercings. A paired samples t-test was done for the two scenarios. There was a statistically significant difference with the means of the two scenarios. Findings, based on two scenarios, showed that people with facial piercings are viewed as less capable of doing quality work, less trusting, less knowledgeable, less engaging, less respected, and they would overall by hired less often. |
| Abstract # 16 |
| John Hiers, Brian Canada PhD, Torpedoes at Night: A Science-Fiction Video Game Inspired bythe U.S. Civil War and the Reconstruction Era, Computer Science |
| Evil aliens from the distant future have come back in time to the year 1862 and traveled to Earth with the intention to alter history and stop the successful implementation of Mitchelville—a Freedmen’s Town that briefly flourished on Hilton Head Island, South Carolina—because it helped shape the future for formerly enslaved persons whose ultimate freedom was perceived by the evil aliens as an early threat to their future economic development. A faction of rival “good†aliens, with knowledge of the evil aliens’ plot, have set their minds to stop them. In this game, you take control of one of the good alien pilots, and your mission is to use your advanced weaponry to defeat the evil aliens and preserve the original history of Mitchelville. Torpedoes at Night consists of a player-controlled good alien along with a number of evil aliens that are generated at random. The bad aliens move from right to left, creating the illusion of a moving background. This is a view of the night sky above Earth. The player character moves around the background, at a set speed, evading the bad aliens. The bad aliens move in certain paths and will destroy the player character if touched when the player’s shields are destroyed. The player character can fire at the bad aliens, but the torpedo does not destroy the bad aliens. After the shot is fired the torpedo can be set to explode. If the explosion hits a bad alien it is destroyed. If the counter reaches 30, the bad aliens have been held back. There is also a counter for the strength of the player’s shields. This computes how many times the player character is hit by bad aliens. If the counter reaches 0, he or she is in danger. If the player is hit while the shields are destroyed, it is a failure and the game is over. The game can be played online at https://www.greenfoot.org/scenarios/24932 |
| Abstract # 17 |
| Mikael Nelson, Assessing the Correlation Between Attention-Deficit/ Hyperactivity Disorder and Social Anxiety, Social Sciences |
| You can boldface text! You can italicize text! You can even use superscripts and subscripts! Just make sure your abstract is under 300 words! This study was created to examine the correlation between Attention-Deficit/Hyperactivity Disorder (ADHD) and Social Anxiety. This study was conducted at the University of South Carolina Beaufort. The study reviewed self-reported documents of 50 student participants throughout the university. Previous research has revealed a correlation between ADHD and general anxiety, however, this report will delve into the anxiety of college students from a social aspect to better understand if the level of ADHD that an individual reports is correlated with their social anxiety levels. The results of the study will provide further insight into possible disorders that accompany ADHD. |
| Abstract # 19 |
| Adriana Torrence, Kristy Campos, Brianna O’Brien, and Alex Markle, The Graying of Beaufort County, Social Sciences |
| Our research group began a study that focused on the older adult population in the Lowcountry area. We developed a survey on SurveyMonkey to collect data that would allow us to examine their needs, living arrangements, regular activities, health, and social connections and how they impacted their levels of anxiety/depression and feelings of loneliness. We distributed our survey through Facebook and obtained 58 responses from current Beaufort county residents aged 55 and older. Our main focus was examining the factors that had an impact on respondents who identified as having feelings of loneliness and/or anxiety/depression. Based on our analysis, we found that there is a positive correlation between respondents who identified as participating in fewer social activities and more at-home activities that also described themselves as feeling lonely. Another relationship we were able to examine further was between their social support system and their frequency of how often they see their family and/or friends. Our findings demonstrated a positive correlation between respondents that identified as feeling lonely less often who also see their family/friends more often than the average respondent. In addition to these findings, we also examined those respondents who perceived themselves as anxious and/or depressed. We found a negative correlation between the respondents who perceived themselves as anxious and/or depressed and to having someone to listen when they need to talk and to do something enjoyable with. These respondents had someone, whether a family member or a friend, more often than the average respondent, yet, they had identified as having feelings of anxiety/depression more than those who had stated they had no one to talk to/do something enjoyable with. Those respondents who identified themselves as anxious/depressed also indicated higher levels of loneliness, reported having little to no social engagement in social activities, were unemployed, and currently live alone. |
| Abstract # 21 |
| James Smith, Brian Canada PhD, Between the Lines: A Video Game Inspired by the 13th Amendment to the United States Constitution, Computer Science |
| Between the Lines is a twist on the classic "catch the falling objects" game in that it has the potential to indirectly educate the player about history. In our CSCI B145 class (Java Programming I) from Fall 2019, we were required to develop a video game based on the Reconstruction Era. I decided to focus this game on the 13th Amendment because it was a huge turning point in history in that it established that slavery wasn’t legal. This topic has the potential to interest a wide audience because this Amendment was one of three that helped rid the United States of slavery. As the game progresses, the words of the 13th Amendment will fall from the top of the screen in the order in which they appear in the amendment. Some of these words will be misspelled, and you will control a "typewriter guide" that will catch the misspelled words. If you make it through the entire 13th Amendment, the sequence restarts with the beginning of the amendment, but a different set of words will be misspelled. If you fail to catch any of the misspelled words then the game will end. While overt education of the player is not necessarily a requirement of the game, it is hoped that the repeated re-reading of the 13th Amendment may promote tangential learning in that the player would become more familiar with the meaning of the amendment and that they may want to explore the historical context that led the amendment to be ratified. Try the game online, right in your browser: https://www.greenfoot.org/scenarios/24939 |
| Abstract # 22 |
| Bruce Brassuer, Dr. Ronald Erdei, Using Evolved Evaluation Functions to Generate Cost Functions Optimized for Specific Deep Learning Tasks, Computer Science |
| All deep learning models have a cost function that is used to measure the performance of the model on a given task. This performance value is used to adjust the parameters of the model to increase its performance. Common methods of choosing a cost function for a model are testing multiple human-designed cost functions or using a genetic algorithm to choose among multiple human-designed cost functions. This poster presents a new method in which genetic programming —a technique of evolving computer programs— is used to generate cost functions optimized for specific deep learning tasks. Using this method, a cost function was generated within a single hour that outperformed human-designed cost functions like squared error and categorical cross entropy on small 3-4 layered neural networks when tested on the MNIST data set (commonly used for bench-marking the performance of classification models). This method is not limited to cost functions; it can also generate evaluation functions for other algorithms like genetic algorithms and genetic programming algorithms. |
| (Total 18 abstracts) |