Streamlining R&D for Accelerating Drug Discovery

Situation

A Fortune 500 pharmaceutical company was struggling with the manual creation of clinical concept sets for its R&D processes. The cohort sets are crucial for clinical trials and patient segmentation, yet creating them required extensive manual effort, resulting in inefficiencies and delays.
How could the client accelerate the cohort creation process while minimizing human error and improving operational efficiencies?

Challenge

The manual process of creating clinical concept sets, such as those used for researching Systemic Lupus Erythematosus (SLE), involved mapping thousands of source codes across various domains (e.g., conditions like diabetes, drugs like metformin, measurements like body weight, and procedures like blood tests). This process demanded extensive validation against specifications, complex data cleaning, and precise inclusion/exclusion criteria, leading to a time-consuming, error-prone process that took an estimated 240 hours for each concept set.

Approach

Mu Sigma partnered to develop an automated tool to streamline concept set creation within the client’s framework. Key functionalities included:

• Input Automation: Users could easily upload files containing codes and data to initiate the automation process.
• Automated Concept Mapping: Our tool processed input data by cleaning it, removing duplicates, and standardizing codes for seamless integration.
• Conflict Management: The tool managed naming conflicts and discrepancies in codes, ensuring accurate, error-free data uploads.
• Batch Processing: By supporting bulk processing, we enabled simultaneous creation of multiple concept sets, significantly reducing time and effort.
• Integration: The tool integrated directly with the client’s system, maintaining standard naming conventions and ensuring smooth, automated uploads of concept sets.

Impact

• Concept set creation time dropped from 240 hours to 45 minutes
• Automation reduced human error in the process, ensuring greater accuracy and consistency
• Our tool proved to be scalable across different clinical concepts and phenotypes