An independent research program linking social determinants of health to cancer outcomes in North Carolina, culminating in an evidence-based state policy proposal (that was part of my Truman application).
This project began as a course data science project and grew into an independent public health research program. I analyzed county-level data across all 100 NC counties, modeling the relationship between social determinants of health (SDOH) and breast, lung, and colorectal cancer incidence and mortality using OLS, Ridge, and LASSO regression.
This finding takes on urgent significance given that 35% of North Carolinians live in rural counties where college graduation rates are 43% lower than urban areas (23.0% vs. 40.4%) — the populations with the highest cancer burden have the least access to the most powerful prevention factor identified in this analysis.
The research was presented as a poster at the Grand Challenges Scholars Program Annual Meeting (February 2026) and formed the foundation of a Truman Scholarship policy proposal for a last-dollar scholarship program targeting first-generation rural students, projected to prevent 9,331 cancer cases over 18 years and return $377M annually against $308M in program costs.
A course-length thesis for PHI 325 that moves from diagnosis to legal critique to a data-driven proposal that connects to my ongoing MEPS research.
Part I constructs a convergence argument across six ethical frameworks — relational, virtue, deontological, pluralist, liberal egalitarian, and utilitarian — to establish that the American healthcare system is not merely imperfect but ethically indefensible, grounded in a 9.9-year life expectancy gap between the most and least advantaged Americans.
Part II turns to the Supreme Court itself, arguing that Washington v. Glucksberg's "traditions test" is a self-defeating methodology: the historical record it invokes is selective and self-contradicting, and a Constitution built to reason forward from principle cannot be governed by reasoning backward from tradition. Drawing on the Philosophers' Brief, it argues that only living constitutionalism can adjudicate liberty interests created by modern medicine.
Part III connects these ideas while drawing on my existing data science work (MEPS) by using gradient-boosted regression (XGBoost) on five years of Medical Expenditure Panel Survey (MEPS) data, to model healthcare expenditure by insurance stability. The MEPS work preliminarily found that publicly insured, coverage-unstable populations are systematically overpredicted for preventive care and underpredicted for emergency care, indicating structural access barriers. From a moral perspective, this finding grounds a proposed 12-month stable insurance mandate, defended as legally feasible under existing ACA precedent and ethically required under all four principles of biomedical ethics. Importantly, this implementation would provide the us healthcare system with a minimally defensive ethically component, addressing concerns raised in Part I while using the methodologies proposed in Part II.
Together, these papers are meant to explore the debate surrounding public health in the US and spark conversation on how machine learning, public policy, and legal scholarship can improve institutions.
Policy analysis, bioethical argument, and engineering research across computational biology, health systems, and clinical technology.
Comprehensive policy analysis evaluating three alternative approaches to sustaining telehealth access for Medicaid recipients following the expiration of federal COVID-era waivers. Synthesizes epidemiological evidence showing telehealth reduces mortality by up to 56% for patients with chronic conditions in rural populations. Integrates federalism analysis, stakeholder mapping, and economic impact assessment to develop a policy recommendation for state-level Medicaid provisions addressing rural healthcare access for 34% of North Carolinians.
Engineering research developing firmware and software for novel clinical electroporation devices. Resolved multithreading security vulnerabilities that could cause unexpected system behavior, implemented fail-safe emulator protocols, and redesigned separate clinician and researcher frontends — reducing clinician interaction from 12 clicks to 3 while adding safety bounds visualization and automated patient reporting.
Investigated the effects of estrogen and progesterone on anterior cruciate ligament fibroblast gene expression using a porcine in vitro model of pre-pubescent female development. Found that progesterone has a greater effect on proteoglycan expression in ACL fibroblasts than estrogen, and that TGF-β1 stimulation significantly upregulates ECM production. Motivated by the observed 3–8× higher ACL injury rate in skeletally immature adolescent female athletes compared to males. Supported by the Comparative Medicine Institute SIRI program.
Poster presenting a gradient-boosted regression (XGBoost) analysis of five years of Medical Expenditure Panel Survey (MEPS) data, using insurance stability as a socioeconomic proxy for healthcare expenditure modeling and a Just Train Twice (JTT) bias-correction framework to reduce worst-group prediction error. Findings on access barriers for coverage-unstable populations directly motivate the 12-month stable insurance mandate proposed in Part III of the bioethics thesis above.