Year: 2016-17

Company: Bristol-Myers Squibb

Liaison(s): Madhu Krishnan, Anthony Savoca

Bristol-Myers Squibb ( BMS ) is a global BioPharma company firmly focused on its mission to discover, develop, and deliver innovative medicines to patients. Within BMS is a group called Business Insights and Analytics ( BI&A ) which is responsible for enabling bold, strategic decisions driven by data and analysis. While BMS has strong capabilities to analyze its structured data ( quantitative data ), it has yet to build the capability to analyze its abundance of unstructured, qualitative data. Qualitative data yet to be mined for insights includes transcripts, quality documents, customer complaints, call logs, and other documents. This TMP sought to develop a system that automates the mining and contextualization of BMS unstructured data to uncover patterns that can lead to business insights. The BMS TMP team performed extensive secondary research and KOL interviews to determine the best software candidates for qualitative data analysis. The team evaluated software options on a weighted scale based upon three major criteria: ease of use, price, and operational capabilities ( including natural language processing and machine learning abilities ). Four candidates emerged as the best options which were then tested by training the software’s algorithms and bench marking their respective predictive capabilities when coding new, previously unseen documents. The team developed a comprehensive codebook specific to BMS quality, SOP, and customer complaint documents by conducting in-depth interviews with BMS quality engineers and managers to discover key pain points. A proof of concept experiment was subsequently carried out with over 250+ qualitative documents on the NVivo platform.