Life Sciences

Modernizing life science manufacturing

Issue link:

Contents of this Issue


Page 3 of 3

Real-time Inference with AWS Greengrass & AWS Lambda The last reference architecture phase comprehensively details the end-to-end steps, from store, process and analyze, visualize, predict and inference at the edge. For more information on Modernizing Life Science Manufacturing and other ways AWS can help your organization, visit us at: Modernizing Life Science Manufacturing Using AWS Services 1: Store Production data from control systems and 3rd party historians (e.g. OSIsoft PI) can be sent to AWS and run on Amazon EC2. Historic device data can be saved to Amazon S3 for raw storage preparation for follow-on analysis or archived to Amazon Glacier. 2: Process and Analyze Connect historical plant device data hosted on Amazon S3 to AWS Glue for ETL operations that apply a CDM in preparation for follow-on whole floor device analytics. This data can be ingested into a database on AWS or stored as files on Amazon S3 to be used by S3 supported analytics environments like Amazon EMR, Amazon Redshift Spectrum, or Amazon Athena. ETL Processing Analytical Querying via Hadoop/Spark MES/SCADA IoT Data Stream Batch Aggregate AWS Greengrass Drug Manufacturing Plant AWS Raw Materials Blending Realtime & Analytics Dashboard Granulation & Drying Secondary Blending Tablet Press Coating & Packaging Analyst AWS Direct Connect Production Data Processing Deep Learning Data Visualization AWS Glue Jupyter Notebooks Amazon QuickSight AWS EMR Amazon Glacier for Archival and Legal Holds Production Data Processing ML Models IoT Environmental Sensors IoT -CCTV Data Lake Tier-2 Storage: Transformed Data Using Common Data Model Amazon S3 Data Lake Compliant to GAMP Archival standards 1. Store 3. Visualize 2. Process & Analyze 5. Infer @ The Edge Visualization Dashboard Corporate Office VPC Subnet Amazon SageMaker 4. Predict RAW Data Transfer CDM Protocol Conversion AWS IoT/Device Management IoT Rule (All Data) Inference Models built from Sensor Data AWS IoT Analytics 3: Visualize Visualization tools like Amazon QuickSight and Jupyter Notebooks can connect to analytics environments and be supported by ERP dashboards to provide Pharma Manufacturing stakeholders the information they need to optimize their full- scale plant floor operations. 4: Predict Predictions using AI/ML models via Amazon SageMaker on device data that is historical or streaming. 5: Inference at the Edge Machine learning inference models created from historical plant sensor data can detect anomalies in live sensor data at the edge with AWS Greengrass and local AWS Lambda functions and report back to AWS.

Articles in this issue

Links on this page

view archives of Life Sciences - Modernizing life science manufacturing