Research Trajectory
My research trajectory reflects a progression from individual-level clinical care to data-driven population health research. My clinical experience in otolaryngology provided the foundation for understanding disease, diagnosis, treatment, and patient follow-up at the individual level. This clinical perspective gradually expanded toward a population-health approach focused on understanding disease patterns, social and environmental determinants, disparities, and opportunities for prevention.
My early public health research focused on community-level health needs, including food access and obesity. Using survey-based primary data and community assessments, I examined patterns of food availability, socioeconomic characteristics, and health-related needs. These experiences introduced me to needs assessment, descriptive epidemiology, and community-based intervention planning, while strengthening my use of R, Excel, and geographic and environmental assessment tools.
My research subsequently expanded into the analysis of large secondary health datasets, particularly the All of Us Research Program and NHANES. In the All of Us data, I examined mental health, metabolic conditions, cancer treatment, and sociodemographic characteristics among women with breast cancer. Multinomial regression was used to evaluate associations between mental health profiles and metabolic, treatment, and sociodemographic characteristics. I then extended this work from conventional regression toward unsupervised machine learning, using K-means clustering to identify distinct phenotypic patterns based on co-occurring mental health, metabolic, treatment, and sociodemographic characteristics.
In parallel, my NHANES research developed my expertise in complex survey analysis and Bayesian modeling. I used survey-weighted regression to account for the population-based sampling design and subsequently applied Bayesian logistic regression to incorporate prior information and quantify uncertainty through posterior inference. This work incorporated R, survey, brms, and Stan, along with model diagnostics and MCMC-based inference.
Together, these projects represent a progression from describing health problems to modeling risk, discovering phenotypes, and identifying priority populations. My current methodological direction integrates epidemiology, biostatistics, Bayesian methods, machine learning, and health data science to move beyond identifying associations toward actionable population-health insights.
Ultimately, my research trajectory is oriented toward translating data-driven evidence into needs assessment, priority-setting, intervention planning, grant development, implementation, and evaluation. The long-term goal is to connect clinical knowledge with population-level analytics so that health interventions can be targeted to populations with the greatest needs and evaluated for measurable impact.
Clinical Care → Public Health → Community Needs Assessment → Secondary/Biomedical Data → Regression → Survey Methods → Bayesian Modeling → Machine Learning → Phenotype Discovery → Priority Populations → Intervention Planning → Implementation → Evaluation → Population Health Impact