Project Title: Using Unassigned NMR Chemical Shifts To Model RNA Secondary Structure

Neel Moudgal, 17, of Saline, MI, developed a computer model that can more easily predict the structure of RNA molecules than current methods that require expensive equipment and a lot of computer power. Neel simplified the process by creating a library of possible structures for a given RNA molecule. This allowed him to rapidly and reliably predict the structure of various RNA molecules using only easily accessible data. He hopes his work will make the diagnosis and treatment of diseases easier in the future.

• Photos of Neel
• Winner Video
• Bio & Project page
• B-Roll

Project Title: Demographic Correlates of Humanizing Language in Media Coverage of Crime: Evidence From The Boston Globe, 1976-84

Emily Ocasio, 18, of Falls Church, VA, tracked homicide victims’ demographics (race, gender and age) in press coverage by The Boston Globe. She examined FBI homicide records between 1976 and 1984 and matched the victims with their corresponding newspaper articles. Then, using an AI program, she assessed each article to measure the amount of humanizing language used to describe the victim. She found the newspaper reported fewer humanizing details for Black male and young Black female murder victims than for whites.

• Photos of Emily
• Winner Video
• Bio & Project page
• B-Roll

Project Title: A Deep Convolutional Neural Network for Kawasaki Disease Diagnosis

Ellen Xu, 17, of San Diego, CA, developed a computational model that uses deep learning to improve diagnosis of Kawasaki disease (KD), the leading cause of acquired heart disease in children one to five. KD diagnosis is based on five visual signs easily confused for other diseases. Using images of patients with either KD or look-alike diseases, Ellen pre-trained her model and created a neural network that can evaluate a smartphone photo of a child and predict, with 85% specificity, the probability of the child having Kawasaki disease.

• Photos of Ellen
• Winner Video
• Bio & Project page
• B-Roll

Project Title: A Generalization of q-Calculus Using Formal Group Laws

Max Misterka, 16, of Harrisonburg, VA, studied q-calculus, a version of traditional calculus, and generalized it to what he calls s-calculus. Initially, Max set out to prove properties about a related topic concerning F-de Rham complexes but had a key insight and introduced a new tool that he called the s-derivative. With his generalization, Max was able to prove many things about s-calculus which are analogous to known results in calculus and q-calculus. He hopes his new tools will prove useful in quantum physics.

• Photos of Max
• Bio & Project page
• B-Roll

Project Title: Thyroid Hormone Treatment of Neurological Trauma Using the Model Galleria mellonella

Linden Chi James, 17, of Durham, NC, conducted research to see if a thyroid hormone, T3, could potentially be used to treat traumatic brain injury (TBI). They used the wax moth caterpillar as their model because it shares neurological similarities with humans and has a version of human T3 called juvenile hormone (JH). Linden inflicted caterpillars with TBI and treated them with either JH or a control substance. Based on caterpillar agility tests before and after treatment, they believe T3 may be a promising treatment for TBI.

• Photos of Linden
• Bio & Project page
• B-Roll

Project Title: Design of a Novel, Dual-Functioning, Tissue Plasminogen Activator and Factor XI-Inhibiting Anticoagulant Therapeutic for Rapid Ischemic Stroke Treatment

Ambika Grover, 17, of Riverside, CT, created a targeted injectable microbubble designed to break up blood clots in the brains of stroke victims and restore blood flow. Ambika used magnetic iron oxide nanoparticles, which are so small they can only be seen with a special microscope, coated with a layer of a drug that breaks up clots and a layer of a drug that prevents new clots from forming. Then she enclosed the nanoparticles in a shell to create a microbubble and coated it with proteins called peptides, which are attracted to blood clots.

• Photos of Ambika
• Bio & Project page
• B-Roll

Project Title: Online Learning of Smooth Functions

Ethan Zhou, 18, of Vienna, VA, studied the mathematical theory behind a type of machine learning in which the program receives training data gradually. His work should be especially useful for algorithms that predict events that are naturally revealed over time, such as weather. For example, the program may be tasked with predicting Tuesday’s weather and then Wednesday’s weather after it learns if Tuesday’s forecast was correct. Ethan investigated how well these algorithms do when confronted with something very unpredictable.

• Photos of Ethan
• Bio & Project page
• B-Roll

Project Title: Selective Inhibition of Matrix Metalloproteinase-9 Attenuates Traumatic Brain Injury Mediated Blood-Brain Barrier Disruption in a Novel Dynamic in vitro Model

Samantha Maya Milewicz, 17, of Armonk, NY, studied how the body’s immune reaction to traumatic brain injury (TBI) leads to secondary injury by causing the blood-brain barrier (BBB), which protects the brain from toxins, to leak. Using a new 3D model of the BBB, Samantha found that after a TBI, the brain overproduces an enzyme, which degrades the proteins that bind together the cells lining the BBB. Reducing the production of that enzyme stopped the leakage, allowing the BBB to function normally.

• Photos of Samantha
• Bio & Project page
• B-Roll

Project Title: SuiSensor: A Novel, Low-Cost Machine Learning System for Real-Time Suicide Risk Identification and Treatment Optimization via Computational Linguistics

Siddhu Pachipala, 18, of The Woodlands, TX, aimed to develop a machine learning tool that assesses suicide risk and best course of treatment based on patients’ diary entries. Based on his models, correlations between semantics (meaning), suicide risk and course of treatment were more accurate than correlations between syntax (grammar), risk and course of treatment. He believes using semantics to gauge psychological health is worth future study, as current measures have led to under-detection of suicide risk.

• Photos of Siddhu
• Bio & Project page
• B-Roll

Project Title: QPOML: Leveraging Machine Learning and Game Theory To Detect and Characterize Quasi-Periodic Oscillations in X-Ray Binaries

Thaddaeus Kiker, 18, of Fullerton, CA, predicted the presence and properties of quasi-periodic oscillations (QPOs) – a mysterious harmonic flickering in the x-ray radiation from black holes. Although QPOs have been observed for decades, scientists have not yet agreed on a definitive explanation for them. Using 16 years of data from two black holes, Thaddaeus trained new machine-learning models to predict the presence and properties of their QPOs. He hopes his work will help solve some of the mysteries that remain.

• Photos of Thaddaeus
• Bio & Project page
• B-Roll

• Seaborg Speech