Miriam Osei-Bonsu

Profile

Senior statistician with 9 years of experience designing experiments and building predictive models in pharmaceutical and public-health settings. Specializes in mixed-effects models, survival analysis, and regulatory submissions under FDA guidelines. Reduced model validation cycle time by 38% at Veridian Biopharma through automated SAS macro libraries adopted across three study teams.

Work Experience

Senior StatisticianMar 2020 - Present

Veridian Biopharma, Durham, NC

Designed and analyzed Phase II/III adaptive clinical trials for two oncology compounds, reducing sample-size over-runs by 22% through pre-specified interim analyses using SAS PROC MIXED and R survival packages.
Built an automated SAS macro library for TLF generation that cut statistical programming time by 38% and was adopted by 3 additional study teams within 6 months of release.
Collaborated with biostatistics leads to prepare 4 regulatory submission packages (NDA/BLA) meeting FDA 21 CFR Part 11 standards, with zero major deficiencies cited during review.

Statistician IIJun 2016 - Feb 2020

Meridian Health Analytics, Raleigh, NC

Developed Bayesian hierarchical models in R to forecast regional hospital readmission rates across 47 facilities, improving 30-day readmission prediction accuracy by 17 percentage points versus legacy logistic regression.
Conducted power analyses and sample-size calculations for 12 observational studies, ensuring 90% statistical power while keeping recruitment budgets within a combined $2.1M constraint.
Mentored 2 junior statisticians in mixed-effects modeling and reproducible reporting with R Markdown, cutting peer-review revision cycles from 3 rounds to 1 on average.

Statistical AnalystAug 2014 - May 2016

Carlow Survey Research Group, Charlotte, NC

Applied complex survey weighting (raking, post-stratification) to 8 statewide public-opinion datasets totaling 94,000 respondents, bringing design-effect-adjusted standard errors in line with AAPOR standards.
Automated monthly report generation in Python and pandas, eliminating 12 hours of manual data-cleaning per reporting cycle.
Built a dashboard in Tableau connected to a PostgreSQL back-end that surfaced key demographic cross-tabs for 5 client accounts, reducing ad-hoc analyst requests by 40%.

Profile

Work Experience

Education

Skills

Links