Our client is one of the world’s largest pharmaceutical company producing medicines and vaccines for a wide range of medical disciplines, including immunology, oncology, cardiology, endocrinology, and neurology.
The client was facing the challenge of different demand patterns of drug combinations for each country. This inconsistent demand prediction system that was modified by local sales teams based on their experience and market knowledge provided below par results.
The client was looking for a solution that would enable them to achieve seamless and faster deployment of new changes and models into production.
We developed an MLOps-based solution for automated ML lifecycle through an automatic model selection framework that would identify the best model combination for each demand category. The automatic model selection framework would
Our ML pipeline enabled rapid pushing of new approaches into production, and achieved the following winning outcomes for the pharmaceutical giant:
15% reduction in forecast error (root mean square error)
Reduced costs by ~2.5%
Boosted sales revenue by 1.5% (~$600 million)