Imran S.

I am a PhD statistical consultant helping researchers, clinical teams, academic authors, and organizations solve complex statistical problems and turn quantitative evidence into defensible, publication-ready or decision-ready conclusions. My work goes beyond running standard statistical tests. I align the research question, data structure, analytical assumptions, model choice, validation strategy, uncertainty, and interpretation so that conclusions are statistically sound, reproducible, and defensible under peer review or stakeholder scrutiny. I also support the statistical evaluation of AI and predictive systems. This includes performance-metric selection, discrimination and calibration, uncertainty quantification, validation-set sample-size and precision reasoning, subgroup performance, robustness and sensitivity analysis, and external validation and generalizability assessment. CORE EXPERTISE • Causal inference for observational and real-world evidence studies • Statistical evaluation of AI and predictive systems: performance, calibration, uncertainty, subgroup analysis, robustness, and generalizability • Survival analysis, Cox models, and competing risks • Longitudinal and mixed-effects modeling • Bayesian hierarchical modeling and evidence synthesis • Clinical and RWE predictive modeling and validation • Cost-effectiveness analysis, ICER/QALY interpretation, and probabilistic sensitivity analysis • SEM, CFA, mediation, and psychometric validation • LPA/LCA, mixture modeling, covariates, and distal outcomes • Regression, logistic, panel, and applied econometric modeling • Statistical manuscript support, Methods/Results development, and reviewer-response support • Monte Carlo simulation, robustness checks, and sensitivity analysis SELECTED CLIENT WORK • Completed a 5-star Latent Profile Analysis project involving covariates and distal outcomes, including model comparison, R3STEP and BCH-style interpretation, publication-ready tables, and reviewer-facing explanations. SELECTED PORTFOLIO WORK • Clinical/RWE causal-inference workflow using IPTW, doubly robust AIPW estimation, covariate-balance assessment, overlap diagnostics, and sensitivity analysis. • HEOR/HTA-style probabilistic cost-effectiveness model using expected costs and QALYs, ICER, net monetary benefit, 10,000 Monte Carlo simulations, and scenario analysis. • Bayesian hierarchical evidence-synthesis workflow across 22 clinical trials, including prior and posterior predictive checks, convergence diagnostics, heterogeneity assessment, future-trial prediction, and prior-sensitivity analysis. • Clinical prediction and statistical validation workflow for 30-day diabetes readmission, including leakage prevention, protected test evaluation, benchmark comparison, probability calibration, feature-importance analysis, and careful interpretation of predictive performance. WHAT YOU WILL RECEIVE • Publication-ready tables and figures • Reproducible scripts, syntax, and structured outputs • Clean and documented datasets where applicable • Clear statistical interpretation, not only software output • Manuscript-ready statistical interpretation, methodological explanations, and reviewer-response support • Careful discussion of assumptions, uncertainty, limitations, and robustness • Structured milestone-based delivery for larger projects TOOLS R | Mplus | Python | Stata | SPSS | AMOS SmartPLS available when specifically required. WORKING STYLE Upwork Verified: Detail-Oriented Work Style. I work carefully, communicate clearly, and keep conclusions within the limits of the evidence. My goal is not simply to produce model output, but to ensure that the statistical approach fits the research question and that the results can be understood, reviewed, and defended. For complex, publication-stage, or initially unclear projects, I usually recommend beginning with a focused fixed-price milestone covering the research question, dataset structure, model feasibility, key assumptions, deliverables, and reporting expectations. The main analysis can then proceed once the scope is confirmed. For AI or predictive-system evaluation, please include the intended use, model output, reference standard, evaluation dataset, current performance results, relevant subgroup variables, and validation objective. Please send a brief overview of your research question, dataset or manuscript stage, current statistical or methodological concern, and required deliverables. I will identify the safest first milestone and the most appropriate analytical or publication-support path.