My research lies at the intersection of applied mathematics, biostatistics, and global health. I develop mechanistic and stochastic models of infectious disease transmission using ODEs, SDEs, and PDE-based frameworks where applicable. I integrate these with statistical and machine learning approaches—including decision trees, random forests, neural networks, and XGBoost—to study antimicrobial resistance, hospital-acquired infections, and health disparities. My goal is to build predictive and interpretable tools that inform both scientific understanding and real-world public health decisions.
Modeling Antimicrobial Resistance in a One Health Framework
Developed a combined ODE/SDE modeling framework to analyze the spread of resistant pathogens across animals, the environment, and farmworkers.
Applied next-generation matrix methods to estimate pathogen-specific reproduction numbers.
Evaluated the effectiveness of multiple biosecurity interventions (e.g., animal movement controls, hygiene, and antimicrobial stewardship).
Key result: Targeting host-to-host transmission achieved the greatest reduction in infection across both livestock and human populations.
Machine Learning for Urinary Tract Infections (CDC Fellowship)
Built and evaluated models including Decision Trees, Logistic Regression, Random Forest, Neural Networks, and XGBoost to classify hospital- vs. community-acquired UTIs.
Applied SMOTE and cross-validation to address imbalanced data.
Key result: Random Forest achieved the strongest performance (AUC ≈ 0.94); nurse unit type consistently emerged as the top predictor.
Impact: Contributed to risk stratification and early identification of hospital-acquired infection cases.
Acute Pyelonephritis Risk Factors
Applied logistic regression, survival analysis, and machine learning (RF, XGBoost), combined with propensity score matching (PSM) to analyze a large inpatient dataset.
Key Result: Urinary retention increased the odds of developing acute pyelonephritis more than fivefold (OR ≈ 5.4, 95% CI 4.0–7.4), independent of comorbidities.
Impact: Provides actionable evidence for clinicians to prioritize retention as a high-risk indicator.
Bayesian & Agent-Based Modeling of MRSA
o-authored research applying Bayesian inference to estimate MRSA transmission rates across hospital units and used agent-based models (ABMs) to simulate inter-facility dynamics.
Key Result: Bayesian models revealed unit-specific hotspots (e.g., surgical ICU, inpatient wards), and ABMs showed that limiting shared staff to <5% sharply reduced cross-facility outbreaks.
Impact: These methods inform targeted interventions at both intra- and inter-hospital levels.
SINDy + Uncertainty Quantification (UQ)
I plan to develop a UQ-extended Sparse Identification of Nonlinear Dynamical Systems (SINDy) framework to identify governing equations directly from high-dimensional, noisy clinical data. This approach will allow discovery of mechanistic structures underlying hospital-acquired infection dynamics while incorporating uncertainty in estimation.
Integration of Causal Inference
To better understand the social and clinical drivers of infection risk, I aim to build a causal inference framework using propensity scoring, instrumental variables, and causal forests to disentangle complex relationships between socioeconomic factors, clinical practices, and health outcomes.
Data-Driven Global Health Models
I seek to integrate EHR data, genomic data, and population-level metrics to develop scalable, equity-informed models of AMR spread. These models will support precision public health interventions targeted at vulnerable populations worldwide.