Projects
Please follow along with some samples of my work - both in group and individual settings.
I look forward to working on more projects and keeping you updated!
February 2023
In this project, we aim to leverage the granularity of single-cell data from AD and control samples in an attempt to identify new cell types or genes that may be associated with Alzheimer’s. Since AD is a progressive disease, it is important to study the cell-level alterations, as it is these small changes over time that lead to the full prognosis. We will combine data from multiple brain regions to get a more comprehensive view of how the entire brain is affected. This will provide new potential therapeutic targets and a better understanding of the cellular characterization of the disease.
February 2023
In this project, we use an Alzheimer’s MRI Pre-processed dataset from Kaggle containing 6400 images to predict Alzheimer’s disease state. Using this dataset, we aim to apply a deep learning framework consisting of multiple machine learning models to predict a patient’s Alzheimer’s disease state as Non Demented, Very Mild Demented, Mild Demented, and Moderate demented. Our goal is to compare the three models: a standard convolutional network (ConvNet), a fine-tuned EfficientNetV2 architecture (Pre-trained ConvNet), and a Vision Transformer (ViT), to determine which has the best performance for our prediction classification task, and visualize the best model’s learning using an integrated gradients approach to gain a deeper understanding and explanation for the model’s predictions.
February 2023
In this project, the dataset originated from a research article that was interested in developing new methods for doublet detection (droplets containing two cells) in single cell analysis, as well as determining sample identities in singlets (droplets with a single cell). They evaluated their algorithm, demuxlet, on single cell expression data from eight pooled lupus patients using peripheral blood mononuclear cells (PBMCs). The aims of this project are now to investigate the single cell dataset generated by the research article for further analysis.
November 2021
In this project I aligned raw NGS reads of exome-captured DNA to the reference human genome, variant called to determine where difference between the sequences occurred, and annotated the results using information from various databases. I chose to focus on three different nonsynonymous variants that resulted in maple syrup urine disease, Sandhoff disease, and Parkinson’s disease. Ultimately, there is a lot of potential in sequencing and personal genomics to positively contribute to diagnostics and disease risk mitigation within a clinical setting. However, more research must be performed and more policies written for ethical, legal, and social grounding before personal genomics becomes commercially available.
March 2021
​In this project I investigated what affects happiness around the world and how happiness has changed over time. I utilized data from the World Happiness Report which provides a Cantril ladder (happiness) score, along with other features such as logged GDP per capita, social support, healthy life expectancy, freedom to make life choices, generosity, and perceptions of corruption for each country. I compared differences in happiness across world regions, countries, and attempted to uncover which features correlate to an increased happiness score. Read more to find out who is happy and why!
March 2020
The City of Chicago, like other major American metropolitan areas, has long struggled with equity in the provision of acute care services to diverse patient populations. However, whether racial and geographic factors are statistically significant determinants of delays in ambulance response in Chicago remains an open question. The failure of previous research to robustly investigate these connections could pose a barrier to the equitable provision of EMS services in Chicago, as a thorough understanding of the role of social determinants in EMS response times is necessary to best implement policy to address disparities in care. This need has served as the impetus of this study, in which we conducted a multimodal exploratory analysis of an EMS database to investigate the existence of a correlation between demographic and temporal explanatory variables and delays in EMS response.
January 2020
In this group project, my partners and I found a UFO data set from the National UFO Reporting Center, and we were inspired to scientifically analyze the various patterns affiliated with different types of sightings. At the onset of the project, we were interested in analyzing temporal, geographical, and characteristic trends affiliated with each sighting. This included analyzing the location of the sighting, when it was sighted, how long it was seen for, and the shape/characteristics of the sighting. As time progressed, we were also interested in the relationship between real world phenomena and the resulting impact on the number of sightings. We found two key phenomena associated with increased reported sightings, which included the end of the Mayan Calendar (2012) which was believed to be the end of the world and Roswell, NM, famed as the site of an alleged UFO crash in 1947.
Jupyter Notebooks
After I joined a machine learning lab at UChicago, I began experimenting with Jupyter Notebooks and learning different methods and tools for working with various data types. I created this Jupyterbook to showcase a few of my Jupyter Notebooks that I had worked on. These notebooks include linear regression and PCA, data cleaning and investigation, A/B testing, fuzzy matching, convolutions and filtering, and gradient optimization.
