[et_pb_section bb_built=”1″ _builder_version=”3.0.47″][et_pb_row _builder_version=”3.0.47″ background_size=”initial” background_position=”top_left” background_repeat=”repeat”][et_pb_column type=”4_4″][et_pb_text _builder_version=”3.0.86″ background_size=”initial” background_position=”top_left” background_repeat=”repeat” background_layout=”light”]

Congratulations to the 2018 finalists from the University of Utah Science & Engineering Fair! These students will head to the Intel International Science & Engineering Fair in Pittsburgh, PA, May 13-18.

Read below for more information about the finalists and their projects.

[/et_pb_text][/et_pb_column][/et_pb_row][et_pb_row _builder_version=”3.0.47″ background_size=”initial” background_position=”top_left” background_repeat=”repeat”][et_pb_column type=”4_4″][et_pb_toggle title=”Daphne Liu and Mia Yu, West High School: %22Undetected Suicide: Classification of Undetermined Drug Overdose Deaths Using Machine Learning Techniques%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

Suicide is a serious but largely preventable public health problem. Suicide underestimation is common and the issue is particularly severe for drug overdose deaths. Drug overdose deaths are typically classified as intentional (suicide), unintentional (accident), and undetermined (intention unknown). Each year in the state of Utah, there is a high proportion of drug overdose deaths being classified as undetermined due to the complexity in the decision-making processes. The recent advances in machine learning technology provide a new way to solve this type of problem. Using innovative machine learning techniques, this study aims to accurately classify the undetermined cases into either suicides or unintentional deaths. The secondary objective is to present temporal trends in drug overdose suicide rates by age, sex and race in Utah. Utah drug overdose data in year 2012 to 2015 are extracted from the National Violent Death Reporting System (NVDRS). The cases that are already classified as suicides or unintentional in the data are used to train and test our machine learning models. Three classification algorithms, including Random Forest Classifier (RFC), Support Vector Machine (SVM), and Artificial Neural Networks (ANN), are used to build the models of classification. The results show that the overall classification accuracy rates of all three models are 94percent. The SVM model with the best F score of 90 percent accuracy rate for suicide classification is finally selected to classify the undetermined cases into suicide and unintentional deaths. According to findings produced by the SVM model, drug overdose suicide rates are underreported by 34 percent averaged across the four years, ranging from 31 percent in 2013 to 38 percent in 2015. Temporal trends from 2012 to 2015 in drug overdose suicides are presented by age, sex and race subgroups. This research identifies a cost-effective method to substantially reduce suicide underreporting which, if replicated elsewhere and implemented widely, can potentially enhance the quality of suicide surveillance and research and facilitate the development of effective suicide prevention programs.

[/et_pb_toggle][et_pb_toggle title=”Marina Gerton, West High School: %22Mucoadhesive HA-based film releasing metronidazole to treat bacterial vaginosis%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

Bacterial vaginosis is a very prevalent women’s health issue that affects millions of women worldwide every year, and many current treatments are messy, inconvenient, and ineffective. Therefore, I wanted to develop a new method of delivering metronidazole that would be more effective, more convenient, and at a low pH similar to that of the normal vaginal environment. Films were made by crosslinking modified hyaluronic acid, using a crosslinker, to create a hydrogel, in which metronidazole or metronidazole benzoate and methylcellulose were incorporated, and was dried to a thin film. Release testing was run by placing the films in simulated vaginal fluid and assessing drug concentration using UV spectroscopy. Mucoadhesion tests were run by placing strips of film on bovine vaginal tissue and attaching a syringe pump filled with simulated vaginal fluid. A formulation was found that created films that allowed for a controlled release of the drug over multiple days, the proper pH, and adhesion to the tissue for multiple days.

[/et_pb_toggle][et_pb_toggle title=”Divyam Goel, West High School: %22Viruses to the Rescue: Using D. Tsuruhatensis to Demonstrate Phage Therapy’s Effectiveness Against Antibiotic Resistant Bacterial Biofilms%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

Phage therapy is the use of bacteriophages to treat bacterial infections. In this project, The bacterium D. Tsuruhatensis was used as a model organism to demonstrate the effectiveness of a freshly isolated lytic phage against biofilms formed by the bacterium. The phages were tested against a solution of ampicillin and kanamycin. The phage solution showed almost complete eradication of the biofilms whereas the antibiotic solution did a poor job.

[/et_pb_toggle][et_pb_toggle title=”Shriya Pingali, West High School: %22Using Neural Networks to Optimise Key-Length Prediction for Polyalphabetically Encrypted Text%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

With the continuous rise of machine-learning technology, cybersecurity algorithms are faced with a new challenge. This project seeks to examine the efficiency of machine-learning on predicting the key-length used to encrypt text via the Vigenere cipher, a method of encryption which, due to its polyalphabetic nature, serves as the basis for more complex security algorithms used today. The Kasiki method is a generally unreliable means of determining key-length for the Vigenere cipher, but it is so far the only conceived method. This experiment examines the accuracy to which the Kasiki method and neural networks can individually and jointly predict key-length. It was found that while the Kasiki method and basic logistic-regression machine learning model yielded accuracy rates of approximately 55% and 25% respectively, the model optimized to account for inaccuracies in the raw predictions by the Kasiki method was 75% accurate.

[/et_pb_toggle][et_pb_toggle title=”Alexander Cheng, Hillcrest High: %22Automatic Detection of Intravitreal Neovascularization in Retinal Flat Mount Images Using Deep Learning Methods%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

Retinopathy of prematurity (ROP) is a leading cause of childhood blindness. Recent developments have discovered that ROP can be characterized by intravitreous neovascularization (IVNV), which is when blood vessels grow abnormally and proliferate into the vitreous of the eye. IVNV growth can be seen in retinal flat mounts images, which are images of cut open retinas laid flat on a surface. Current methods of detecting IVNV are manual and inefficient, creating a need for a faster and automated system. Such a system would help progress the discovery of new diagnoses and treatments for ROP. This project utilized a fully convolutional neural network (FCNN) to automatically segment the retinal flat mount images to identify IVNV. Cross-validation and pixel accuracy were used to determine the accuracy by comparing the segmented images from the FCNN to predefined ground-truth markings. The results from this project showed that the method created has good potential to be an automatic screening tool to identify IVNV in the future and replace manual identification. Future research will include removing debris that appear to be IVNV from the images, utilizing different FCNNs, and exploring unsupervised deep learning methods to increase accuracy.

[/et_pb_toggle][et_pb_toggle title=”Emma Sun, The Waterford School: %22Externality Framing Effects on Cognition%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

Although suggestions have been made about how positive and negative externality framing can affect different situations, so far there has been no real empirical testing of these predictions. If the hypotheses are true, this phenomenon could be taken into account in multiple different settings, including changing the way governments and advocates describe problems, as well as generally being a useful tool in shaping public opinion. Thus, a fuller understanding of this predicted phenomenon (effects of positive vs negative externality framing), backed with scientific data, would have significant implications for the future. This project explores the different effects positive vs negative externality framings create, specifically looking at the effects on the different solutions people support (incentive vs. punishment and education vs. compulsion), how serious or important people gauge a situation or issue to be, and how willing people are to take individual action. In particular, it was hypothesized that when positive externality framing was used, people would be more likely to offer incentives to do the right thing or support education to help solve the issue, whereas when negative externality framing was used, people would be more likely to propose disincentives (or punishments) or support solutions that rely on compulsion. It was also hypothesized that those who were presented the issue in a positive externality framing would (1) see the issue as less serious and (2) be more willing to take individual action in comparison to those who viewed the issue in a negative externality framing. To test this hypothesis, two different surveys were administered to approximately 1,000 people (511 received the positive externality framing survey and 520 received the negative externality framing survey) via the online survey tool Survey Monkey. Both of these surveys focused on the same four societal issues: (a) carpooling, (b) vaccinations, (c) the overuse of antibiotics, and (d) coal plant pollution. The only difference between the two surveys was the way the issues were framed. One survey focused on the benefits fixing the societal issue would create for the community (positive externality framing), while the other survey focused on the harms not solving the societal issue inflicted on the community (negative externality framing). Analysis of the collected data showed two strong relationships that supported the hypothesis that positive externality framings made people more likely to propose incentives and education to solve the identified problem, whereas the native externality framings caused people to focus on punishments and compulsions. Using a significance level a = 0.05 and the x2 test for independence, in the carpooling scenario, there was a statistically significant relationship between the positive externality framing and supporting education rather than compulsion as a potential solution, and between the negative externality framing and supporting compulsion over education (x2 = 4.43, p = .0354). Similarly, in the antibiotics scenario, there was a strong relationship between the positive externality framing and supporting a monetary incentive for proper antibiotic use, and between the negative externality framing and supporting a monetary punishment for improper antibiotic use. While this result was not quite statistically significant at the 95 percent confidence level (p = .0574), the relationship is still considered statistically significant at the 90 percent confidence level. Most of the other observed results were also in the expected direction but were not statistically significant. No statistically significant results were found to support the remaining hypotheses about positive externality framings causing people to view the issue as less serious or to be more willing to take individual action.

[/et_pb_toggle][et_pb_text _builder_version=”3.0.86″ background_layout=”light”]

Observers

[/et_pb_text][et_pb_toggle title=”Rachel Maxfield, American Preparatory Academy: %22Creating a Low-cost Solution to Water Problems in Developing Countries%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

“If 90 school buses filled with kindergartners were to crash every day, with no survivors, the world would take notice. But this is precisely what happens every single day because of poor water, sanitation and hygiene.” The United Nations Children’s Fund (UNICEF) once said this in an attempt to bring recognition to how much this problem is neglected compared to others around the world. This project will outline the best way to pump and transfer water in less developed countries. This project is the second stage in creating a device that pumps, purifies, and transports water. Last year the first stage of this project was completed. A trailer was created using recycled bike parts. Before the beginning of this project, much background research was done. Research was done on the different methods of pumping water. This project was commenced with the designing of many different pumps. After many designs were created the best was selected and created. The result was a fully functioning prototype of a reverse peristaltic pump functioning with the pedaling of the pedals as a source of power.

[/et_pb_toggle][et_pb_toggle title=”David Zhong, Skyline High: %22The Impact of Xylem Cavitation on Hydraulic Conductivity%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

Xylem cavitation is the phenomenon whereby air bubbles are formed in the xylem conduits, which can be triggered by water stress (soil drought) or winter-freezing of plants. Cavitation reduces a plant’s capacity to transport water from soil to leaves. The purpose of this study is to understand how cavitation affects the hydraulic conductivity in stems. Water birch stems were gathered and original hydraulic conductivity was measured using a hydraulic conductivity apparatus. Next, negative tension via a centrifugal device was exerted on the same stem to introduce xylem cavitation. The hydraulic conductivity was measured and compared to the original hydraulic conductivity. Each experimental group consisted of seven stems and four different tensions (-2.0 MPa, -2.5 MPa 3.0 MPa, and -3.5 MPa). We found that the hydraulic conductivity decreased as the negative tension increased, which indicates that the cavitation in xylem conduits impairs the ability for water delivery. Gaining a better understanding of how water conductivity works allows for better selection of drought resistant plants and subsequent conservation of water.

[/et_pb_toggle][et_pb_toggle title=”Tarun Martheswaran, The Waterford School: %22Exploring the Transmission and Strategic Intervention of Dengue Fever Using SIR Compartment Mathematical Modeling and Ordinatry Differential Equations%22″ _builder_version=”3.0.86″ open=”off” title_line_height=”1.4em”]

Project Abstract:

This study aimed to test the effectiveness of a novel intervention strategy for Dengue Fever using Ordinary Differential Equations and the SIR Compartment Model, a mathematical epidemiological model for the flow of infectious diseases. With its prevalence having multiplied by more than thirty times in the last fifty years and infecting more than 400 million individuals worldwide each year, Dengue Fever is a pressing global epidemic. In a time when there is no clear cure for Dengue, mathematical modeling serves as a method to understand how various intervention strategies affect the flow of the disease through the SIR compartments: Susceptible Humans, Infectious Mosquitoes, and Infected Humans. Previous epidemiological research has used the SIR Compartment Model to test the effectiveness of three intervention strategies for Dengue Fever: Varying Mosquito Population, Varying Percentage of Vaccinated Human Population, and Varying the Daily Bite Rate of Infectious Mosquitoes. In this study, a new intervention strategy of Varying Transmission Probability is introduced and modeled. Studying varying transmission probabilities is significant due to the factors that are associated with it including climate, global warming, population size, etc. Starting with a population of 0, at a 0.75 probability of transmission, the infectious mosquito population reaches more than 32,000 in 10 days compared to just 18,000 at a 0.45 probability. It was determined that varying the Transmission Probability is a slightly more effective intervention strategy than the other three, as it results in a more contained infectious mosquito population. Mathematical Modeling using Ordinary Differential Equations, the SIR Compartment Model, and R Software allows for the understanding of not only whether or not an intervention strategy is effective, but why it is effective and the extent to which it is effective, thus taking one step closer towards prevention. This research can be expanded upon by testing other intervention strategies and, as well as by applying this study to other infectious diseases that are transmitted from a mosquito to a human, such as the Zika and the Malaria Virus.

[/et_pb_toggle][/et_pb_column][/et_pb_row][/et_pb_section]