How a medical device manufacturer saved millions with AI-driven quality control

Challenge: Scaling manufacturing while controlling quality costs

Our client formerly manufactured drug delivery systems and devices primarily in low-cost areas. They've made a strategic decision to relocate one of the new main products closer to their headquarters to ensure the highest quality standards. With an expected increase in production output in the coming years, it was inevitable to reduce the number of critical quality test dimensions in order to keep these quality control costs low, relative to the increased production output.

How we helped

We developed a data-driven definition of critical dimension to apply across the device platform. Using Random Forecast Algorithms, Principal Component Analysis, and Recursive Feature Elimination, we defined a scalable and reusable decision process for determining the most critical quality control dimensions, removing “uninformative” dimensions. The results were presented in a report including a model to predict final badge performance.

The impact: Cost savings, standardized detection, better data management

• Eliminated 21(out of 103) uninformative quality control dimensions, saving millions annually.
• Implemented AI techniques to standardize failure detection across all devices
• Provided recommendations for data acquisition and management to enhance data capabilities.

‍