Context: TCS for Roche Pharmaceuticals – clinical trial analytics.
Challenge: Manual fraud detection across 2,000+ studies and 35,000 sites. Needed scalable statistical frameworks for regulatory submission and patient safety.
Solution: Led Statistical Monitoring teams, developed JMP-equivalent R Shiny dashboards. Applied Bayesian methods, Monte Carlo simulations, and survival analysis (Kaplan-Meier, Cox PH, Random Survival Forests) to flag anomalies. Built R packages with golem architecture and automated deployment with GitHub Actions.
Outcome: Findings directly informed regulatory submissions and patient safety decisions. Standardized fraud detection across the global clinical trial program.