Academic Projects

CPSC 483 – Leveraging GNNs to Improve AI-ECG Disease Detection for LVSD

• Developed a Graph Neural Network (GNN) framework to enhance AI-ECG–based prediction of Left Ventricular Systolic Dysfunction (LVSD).
• Modeled each patient’s ECGs as graph nodes connected by temporal and demographic relationships to improve classification robustness and interpretability.
• Focused on scalability, noise resistance, and real-world clinical data integration.

---

CPSC 486 – A Mathematical Analysis of Diffusion Processes and GANs for Image Generation

• Conducted a theoretical study of diffusion-based generative models and GAN dynamics using probability and stochastic analysis.
• Implemented and trained diffusion models in PyTorch to empirically validate mathematical findings.
• Compared convergence behavior, noise schedules, and sample quality between diffusion and adversarial frameworks.

---

SDS 230 – Statistical Analysis of Video Game Sales and Ratings (1980–2020)

• Conducted an extensive statistical analysis of global video game sales and rating data spanning four decades (1980–2020).
• Utilized a Kaggle dataset containing over 6,900 entries, performing data cleaning, variable recoding, and transformation for use in inferential models.
• Applied statistical techniques including t-tests, ANOVA, ANCOVA, and multiple regression to identify key predictors of sales performance across markets and platforms.
• Explored relationships between critic and user ratings, genre, and console type, constructing a final generalized linear model to predict global sales outcomes.
• Implemented all analyses and visualizations in R, leveraging libraries such as GGally, car, and stats for correlation analysis, bootstrapping, and stepwise model selection.