Xin Liu

- Supporting non-standard oncology roll-over studies on pediatric and adult patients. - Supporting publications based on a open platform study.

- Supported a feasibility randomized controlled trial of a mHealth App vs. Booklets for sickle cell disease Patient-Facing Guidelines - Identified factors for sickle cell disease knowledge improvement and pain reduction.

- Developed an oncology Phase 1a/b adaptive trial design with dose optimization based on Bayesian logistic regression models (BLRM), which jointly model drug toxicity and efficacy. - Proposed new “decision indices” for drug efficacy-toxicity trade-off quantification. - Conducted simulation trials by parallel computing on high-speed servers, showing the design achieves 70% higher rate of choosing correct optimal doses and 20% fewer samples than MTD finding BLRM designs. - Reduced server job submission time by 80% through writing a tutorial to initiate batch file usage. - Built an R package for oneline implementation of the proposed trial design inside Pfizer.

-Reviewed protocols and statistical analysis plans of 79 clinical trials from ClinicalTrials.gov on their statistical justifications for sample sizes in expansion cohorts. -Compared performance of different mixed-distance dissimilarity clustering methods through biologically meaningful simulations and large clinical dataset --- MIMIC-III for best practices for clustering clinical data.

-Tutored graduate students in 6210 and 6211(Design and Analysis of Studies in the Health Sciences I and II) -Invigilated and graded examinations and homework, corrected exam answers, and held office hours.

Bayesian inference and computation, clinical trials