PhD Statistical Consultant | Advanced Quantitative Modeling (R, Mplus)
I help researchers, academic authors, doctoral candidates, and organizations turn complex data into statistically sound, interpretable, and publication-ready results.
My work focuses on selecting defensible analytical strategies, evaluating assumptions carefully, and producing results that can hold under peer review, supervisor review, or stakeholder scrutiny.
- Latent Profile Analysis (LPA) and Latent Class Analysis (LCA)
- Structural Equation Modeling (SEM) and Confirmatory Factor Analysis (CFA)
- Mixture modeling, covariates, and distal outcomes
- Causal inference and observational-study interpretation
- Clinical, real-world evidence (RWE), and health-research modeling
- HEOR/HTA, cost-effectiveness analysis, and uncertainty-based decision modeling
- Econometrics, simulation, and statistical validation
- Statistical tools: R, Mplus, Python, Stata, SPSS
- Applied strengths: Reproducible statistical computing, interpretable predictive modeling, validation, and simulation-based decision analysis
A reproducible clinical/RWE predictive-modeling workflow emphasizing patient-level leakage prevention, protected test-set evaluation, benchmark comparison, probability calibration, and clinically cautious risk interpretation.
A transparent HEOR/HTA decision-modeling workflow comparing two strategies through expected costs and QALYs, incremental analysis, net monetary benefit, Monte Carlo probabilistic sensitivity analysis, a cost-effectiveness plane, a CEAC, scenario analysis, and reproducible exports.
A reproducible clinical/RWE causal-inference workflow using observational data, stabilized IPTW, doubly robust AIPW, overlap diagnostics, covariate-balance assessment, bootstrap uncertainty analysis, and sensitivity checks.
A reproducible clinical time-to-event workflow using Kaplan–Meier estimation, Cox proportional-hazards modeling, Schoenfeld-residual diagnostics, covariate-standardized survival curves, Aalen–Johansen cumulative incidence, cause-specific hazard modeling, and competing-risk sensitivity analysis.
A reproducible longitudinal clinical modeling workflow using repeated serum bilirubin measurements from 312 patients. The project includes data-integrity auditing, patient-specific random intercepts and slopes, quadratic time modeling, adjusted trajectory prediction, residual diagnostics, and focused sensitivity analyses.
A reproducible Bayesian hierarchical evidence-synthesis workflow using mortality outcomes from 22 randomized clinical trials comparing beta-blocker therapy with control after myocardial infarction. The project includes conventional meta-analysis benchmarks, arm-level binomial modeling, prior-predictive and posterior-predictive checks, convergence diagnostics, between-study heterogeneity assessment, partially pooled trial-specific effects, future comparable-trial prediction, and prior-sensitivity analysis.