For more information on Pharma and Biotech or other ways AWS can help your organization visit us at: aws.amazon.com/health/biotech-pharma
© 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Implementing predictive analytics and insight generation
AWS reference architecture
Analytics: Analytics allows to rapidly develop and deploy models to automate manufacturing process
steps to support either real-time or retrospectively workflows specific to shop floor activity. Event
management from these models can be visualized through a control-tower dashboard which aggregates
all shop floor events into a common multi-device supported view. The primary technical outcome
to support analytics includes the use of standard, scalable services to support ETL, query and model
development. Models are then deployed either at the edge to support stream-based process steps or
in the cloud to handle latency tolerant process manufacturing state changes such as alerts, alarms or
notification events.
3
Connected Worker
Bio Reactor/
Unit Operation
Cameras
ML Inference
ML Inference
Deploy a SageMaker Endpoint
Train Model
S3 Event
Data
Ingest
OFC-UA/Modbus
MQTT
OPC-UA
OPC-DA
Ethernet/
IP
PLC/
DCS
Amazon Kinesis
Data Firehoes
Lambda function
AWS IoT
Greengrass
AWS IoT Greengrass
Connectors
AWS DataSync Agent Historia
Local Storage MES AWS Storage
Gateway
AWS Snowball Edge
AWS Cloud
Data
Lake
Batch Inference
Container
Management
Factory
AWS DataSync
AWS IoT SiteWise
Connector
Protocol Conversion
Amazon S3
(Raw Data)
Amazon S3
(Raw Data)
Amazon
EMR
Amazon
Elastic
Container
Service
Amazon Glue
(ETL Job)
Amazon Sagemaker
Amazon
Sagemaker
Jupyter
Notebook
Amazon Elastic
Container
Registry
Docker Image
Train
AWS Lambda
Amazon SNS
Notification
of predication
result
Amazon DynamoDB