Learn from AWS Healthcare solution architects how to ingest data from live streams (like glucometers) and a data extract source. Use these to build a patient cohort, based on various criteria like lab values, demographics, and other clinical events. From that cohort, you can learn how to run predictive models using machine learning on AWS.
For the workshop, you will be divided into groups of 5-7 participants. We will be using an AWS IOT button to act as a glucometer that sends lab readings to the cloud. Then use AWS Glue to merge these readings with existing clinical records and transform the dataset for machine learning. Finally, we will use a machine learning model running on AWS deep learning AMI to generate predictions.
To make sure you get the maximum value out of this workshop and enjoy building on AWS, we wanted to share with you a few prerequisites that will help you get up and running quickly on the day of the workshop:
This will be a hands-on session where you will be accessing various AWS services. Make sure you have an AWS account and have access to login as an IAM user with admin privileges. If you do not, we recommend signing up for a new AWS account before you come for the workshop.
We will be using an AWS IOT button during the workshop. If you have an IOT button with you, please get it along. If you don’t, no worries. We will have a few buttons to be shared within the group which the participants can take turns to try out.
Please download a MySQL database client like MySql Workbench to run scripts on the database.
The code for the workshop is written in python. Make sure you have your favorite python editor to view code. You will not be required to write new code, just modify existing code with some parameters.
Important points:
The workshop is intended to provide you with hands-on experience of building solutions on AWS. We will provide you workshop guides that give you details of how to build the various components of the soliton. However, it will be good to have some previous AWS experience so you can work through the steps faster. Please also leverage the knowledge of others in the group during the workshop. There will be assistants ready to help you if you get stuck.
We will time box each section of the workshop so we are able to complete the end to end solution in time. If you are not able to finish a particular section, don’t worry. We will provide you script that will accelerate you to the next step of the solution.
Last and the most important point: we want you to have fun while working together in groups. Please let your assistants know how we can make this experience better for you.
5. The Technical Flow
Clinical
Records
Glucometer
(IoT Button)
AWS IoT ingests lab
values from the
glucometer
Integrated data in
Amazon RDS
AWS Glue for
transforming data
Transformed data
aggregated in Amazon
S3 to build a data lake
Transformed data in
Amazon RDS for
machine learning
Batch
Stream
Deep learning AMI
for training and
testing the machine
learning model
The Machine Learning
Model predicts if a
particular patient has
kidney disease or not