Aizon Improves Pharmaceutical Manufacturing With Artificial Intelligence Using AWS

Overview

Aizon built a GxP-compliant AI platform on AWS to help pharmaceutical manufacturers optimize biomanufacturing processes. Using AWS IoT, AWS Lambda, and Amazon SageMaker, Aizon created digital twins of bioreactors that trigger machine learning operations for process optimization. For a multinational pharmaceutical company producing blood-plasma-based drugs, Aizon built unsupervised and supervised learning models that identified two critical process variables; optimizing them improved yield by double digits, worth potentially tens of millions of dollars. AWS also reduced Aizon's compute costs by up to 80 percent.

The challenge

Pharmaceutical manufacturing is a complex and heavily regulated industry. There are hundreds of decisions and variables in each manufacturing process that must be calibrated correctly to deliver safe, effective drugs to patients. As today's market scales, the demands of production management and process data analysis exceed human capabilities. A multinational pharmaceutical company producing promising drugs based on human blood plasma experienced an unexplained decrease in product yield coupled with an unwelcome increase in process variability over several years.

The solution

Aizon built a GxP-compliant AI platform on AWS that creates 'digital twins' of bioreactors: individual bioreactors are connected to the cloud via AWS IoT, and when conditions change within a bioreactor this triggers AWS Lambda to execute a machine learning operation in Amazon SageMaker. For the affected pharmaceutical customer, Aizon created an unsupervised learning model to identify patterns among plasma composition origins across thousands of drug batches, and a supervised learning model to predict yield for each upcoming batch coupled with real-time dashboarding.

Reported business value

Aizon's analyses revealed that the plasma composition origin explained over half of the process variance and that batch clusterization was critical to yield prediction. The company identified two critical process variables and found that optimizing just those variables was sufficient to improve the yield by double digits, resulting in potential gains of tens of millions of dollars for the pharmaceutical customer. AWS also reduced Aizon's compute costs by up to 80 percent.