A selection of my graduate-level research and data science projects.
Epidemiology Survival Analysis Shiny Dashboard MPH Thesis
This project forms the basis of my MPH thesis and examines cardiovascular risk – venous thromboembolism (VTE), ischemic stroke, and myocardial infarction – among transmasculine individuals receiving gender-affirming hormone therapy, compared to matched cisgender male and female referents.
Data come from the STRONG cohort, an electronic medical record-based matched cohort study across Kaiser Permanente health systems (2006-2024), with approximately 345,000 participants. Using Cox proportional hazards models and cumulative incidence curves, I found that transmasculine individuals face elevated VTE and stroke risk relative to cisgender women, supporting the need for long-term cardiovascular monitoring for patients on testosterone therapy.
The interactive dashboard was built in R using flexdashboard, Shiny, plotly, and DT, and deployed on shinyapps.io. It allows clinicians and researchers to explore cumulative incidence curves, forest plots of adjusted hazard ratios, and log-rank test results by cohort and outcome.
Cancer Epidemiology Health Equity Logistic Regression 2nd Place Award
As a SCREP fellow at Morehouse School of Medicine, I conducted a retrospective cohort study of 300 lung cancer patients at Grady Health System, a major safety-net institution serving a predominantly African American population in Atlanta, Georgia.
Background: Lung cancer is the leading cause of cancer-related mortality in the United States, with African American patients disproportionately affected in terms of incidence, late-stage diagnosis, and reduced access to treatment. Small-cell lung cancer (SCLC), representing 15-20% of cases, carries a significantly worse prognosis than non-small cell lung cancer (NSCLC). This study investigated whether African American patients are more likely to develop aggressive SCLC subtypes, or whether disparities stem primarily from delayed diagnosis.
Methods: Retrospective cohort study of 300 adult patients with histologically confirmed primary lung cancer. Variables included race/ethnicity, histological subtype, TNM staging, treatment history, socioeconomic indicators, and smoking status. Statistical analysis used chi-square tests and multivariable logistic regression.
Key Findings:
Implications: Disparities in lung cancer outcomes among African American patients are driven primarily by delayed diagnosis and structural barriers to early screening – not by tumor biology. Public health efforts should expand lung cancer screening programs and address systemic obstacles to timely diagnosis. The safety-net model at Grady demonstrates that equitable treatment delivery is achievable when institutional barriers are minimized.