Home
Overview
Research assistant work at Columbia Business School (2025–present): using national Medicare claims to measure how pharmaceutical and procedural innovation associate with real-world mortality and value for money.
Methods
- Survival analysis on MarketScan / CMS-scale claims with Weibull and related models
- Weighted vintage methodology tying drug approval years and procedure codes to innovation exposure
- Cost-effectiveness framing comparing newer vs. older therapeutic cohorts
- ETL & SQL on Redivis — cleaning and querying 1TB+ structured claims (6M+ patient records in aggregate analyses)
Tools
Python (pandas, statsmodels ecosystem), SQL, regression and survival modeling, data visualization for health economics audiences.
Motivation
I care about the gap between a promising innovation and one that actually reaches patients — this project quantifies that gap using population-scale outcomes, not just trial endpoints.