Research
Using a Convolution Neural Network to Improve Ensemble Tropical Cyclone Track Forecasts across the Atlantic Basin
Stony Brook University (Simons Summer Research)
This research focuses on enhancing the accuracy of tropical cyclone track forecasts by leveraging the power of Convolutional Neural Networks (CNNs). Traditional forecasting models often face challenges in accurately predicting the along-track position of tropical cyclones, leading to significant uncertainties in the storm's projected path. These along-track errors can result in inadequate warnings and misallocation of resources during storm preparations, potentially endangering lives and property. Developed CNN model addresses these issues by correcting systematic biases, resulting in more accurate and reliable forecasts. This advancement is vital for improving preparedness and response strategies, potentially reducing the devastating impacts of tropical cyclones on vulnerable communities.
A Novel Approach for Predicting Large Wildfires Using Machine Learning towards Environmental Justice via Environmental Remote Sensing and Atmospheric Reanalysis Data across the United States
Independent Research
This research addresses the critical need for accurately predicting large wildfires in the United States using easily available environmental data and a scalable model. Analyzing USDA and NASA MODIS data, six machine learning classification models were tested, with XGBoost performing the best at 90.44% accuracy. The model's use extends to identifying disadvantaged communities vulnerable to large wildfires, aligning with Environmental Justice initiatives. By employing protective measures and prioritizing resource allocation, wildfire safety organizations can mitigate the health and ecological consequences of large wildfires, particularly in socioeconomically disadvantaged areas.
High-Entropy Oxides with Lithium Batteries
Argonne National Laboratory, Exemplary Student Research Program (ESRP)
Recent years have shown a demand for a more ecological and efficient alternative to alkaline batteries. Lithium batteries are a prime candidate, prompting research into electrode materials with the potential to store energy effectively and withstand numerous uses.
​
In this research, after synthesizing the anode material and assembling coin cells, the Advanced Photon Source (APS) at Argonne National Laboratory was utilized to conduct EXAFS (Extended X-ray Absorption Fine Structure) spectroscopy on the research material, as well as to carry out cycling tests at the Illinois Institute of Technology.
MASC AI: A Novel Method for Effective Mosquito Data Classification and Mapping
The GLOBE Program
This study focuses on addressing the threat posed by Anopheles mosquitoes, carriers of malaria, which infects millions and kills hundreds of thousands annually. The GLOBE Mosquito Habitat Mappers (MHM) app gathers data worldwide on mosquito larvae, aiding in population tracking. However, manual classification of larval images is time-consuming. To enhance efficiency, the study proposes a Machine Learning-based logistic regression model for mosquito identification, achieving over 80% accuracy. By mapping detected locations on a global scale, the model aids mosquito ecologists, governments, and public health organizations in effectively combatting the spread of mosquito-borne diseases, providing a valuable tool for disease control and prevention efforts.
The Impact of Genetic Analysis on the Early Detection of Colorectal Cancer
University of Chicago (Mathematical and Computational Research in Biological Sciences)
Although the 5-year survival rate for colorectal cancer is below 10%, it increases to greater than 90% if it is diagnosed early. This research hypothesized that analyzing non-synonymous single nucleotide variants (SNVs) in a patient's exome sequence would be an indicator for high genetic risk of developing colorectal cancer. First, the patient's exome sequence and the reference exome sequence were repeatedly aligned to identify the regions of similarity. The alignment between the two sequences that resulted in the most similar regions (optimal alignment) was selected. Next, research performed variant calling and identified variants in the patient's exome sequence. Finally, the remaining selected variants were annotated with biologically pertinent information and explored for their potential roles in human disease by cross-referencing databases. A variant in the FGFR4 gene, known to cause accelerated cancer progression and tumor cell motility, was found in the patient's exome sequence. Studies suggest that this variant corresponds with an increased risk of colorectal cancer, supporting the usefulness of this procedure in early detection of colorectal cancer. This research was expanded to demonstrate that exome sequencing methods are capable of identifying other genetic variants.
Young Lords Organization of Chicago
National History Day (NHD)
The Young Lords Organization (YLO), one of the most impactful groups involved in the Civil Rights Movement, was a group of Puerto Rican activists from Lincoln Park, Chicago, who strived to create equal opportunities for minorities. In largely Puerto Rican neighborhoods of Chicago, the Young Lords fought gentrification and inequality by changing the way people viewed the effects of urban renewal. This study explores the triumph and tragedy of the Young Lords Organization. Although the Young Lords were viewed as a criminal gang, their triumphs in providing low-cost services such as healthcare, childcare, and free food in the community left an everlasting positive impact. Despite gentrification still tragically existing today, the Young Lords set an example by fighting against it and creating awareness.