Pharma disease forecasting

My client, a Fortune-100 pharmaceutical company, owned one of the best-selling drugs in history, with billions in annual revenue in the US alone. In 2018 and 2019 I led a project to investigate business strategies for their upcoming R&D pipeline, with the goal of maintaining a strong market position against new competitors, maximizing total revenue, and optimizing patient outcomes across their portfolio. I oversaw the project from the initial proposal, through project planning and execution, to maintaining a long-term relationship with the client. I managed and trained a new team of five business and data analysts.

I supervised a literature review, to provide an independent view on the market, and, using SQL, performed a longitudinal analysis of over 10 terabytes of insurance claims and electronic medical record data. After comparing our results to the client’s data reports, we caught and helped reconcile several errors in their methodology and supported them in establishing more stringent data practices.

Based on our data analysis, I mocked-up a map of key patient states and flows using system dynamics principles. Through a series of co-design sessions with the client, we iterated until we had an accurate and easily quantifiable and communicable structure. Previously, various internal client teams had conflicting views and descriptions of the market, making it hard to work together effectively, so developing a common language within the organization allowed for clearer communication, easier collaboration, and more effective planning.

I coded an agent-based model in Java that simulated the unique treatment path of every individual in the US with a related disease. Demographics, treatment history, and market events affected the behavior of each virtual patient, informed by our data analysis. We tested and trained the model results against historic data. Custom reports and visualizations from the result of a simulation run were aggregated into an Excel dashboard.

The model was used as a virtual sandbox to quickly test and understand the effect of business strategies at a previously unattainable level of detail and accuracy. Specific outcomes included allocating a $500 million marketing budget to maximize portfolio success, quantifying and monetizing potential risk and upside to company goals that were shared with stakeholders and investors, and estimating production quantities to meet future demand.

Prior to this project, my employer had a shaky relationship with this client, but with my attention to detail, close collaboration, and ability to understand nuances of a complicated system, I salvaged the relationship and created a solid foundation for future work. Not only was the contract renewed or expanded every year since, continuing to be the largest stream of revenue for my former employer, the client also requested our support on many additional projects.