Merck Improves Speed and Reduces Manual Effort to Assess Change Proposals Up to 70% on AWS

Biopharma companies operate in a complex regulatory environment governed by over 100 global regulatory agencies. This dynamic industry generates thousands of change proposals annually, each of which must be assessed by a chemistry, manufacturing, and control scientist. Currently, scientists must research across multiple repositories and documents to gather information and adjudicate an assessment for regulatory reportability.

Merck used Amazon Web Services (AWS) to build the Change Assessment Knowledge Engine (CAKE). “CAKE helps chemistry, manufacturing, and control scientists make change assessments more quickly, effectively, and consistently,” says John Baker, executive director of machine learning and analytics at Merck.

CAKE is powered by an Amazon Neptune graph database containing information about Merck’s supply chain and regulatory operating environment. When a new change proposal arrives, CAKE applies a natural language processing model trained on historical data to identify the affected product’s markets, relevant reporting requirements, and country regulations. CAKE then generates recommendations for the scientist on regulatory impact reporting, along with accompanying hyperlinks to specific regulations and Merck’s internal repository.

CAKE reduced change assessment duration by up to 90 percent and total manual effort per assessment by 30–70 percent. Merck developed and implemented its CAKE pilot in less than 10 months using AWS, and in 2021, the company hopes to execute a global rollout. To learn more about CAKE, watch Baker’s talk at re:Invent 2020.

“AWS delivers results, plain and simple,” Baker says. “They’re obsessed with delivering long-term customer value. They strive for the highest standards and iterate quickly with a real bias for action.”