CASE STUDY

Optimizing Data Architecture with AWS to Improve Patient Care for Lifescan

Download

INTRODUCTION

Customer
Overview

LifeScan is a US-based medical device company that specializes in diabetic treatment and management.

More than 20 million people around the world depend on its OneTouch^® blood glucose monitoring devices to help them manage their condition. For over 35 years, LifeScan has had an unwavering commitment to improving the quality of life for people with diabetes by developing products defined by simplicity, accuracy, and trust.

specialists in diabetic treatment
20 million people use OneTouch^®
+35 years commitment

Business Goals

CLIENT OBJECTIVES

Higher Service Levels

AUTOMATE & ACCELERATE

Operating Efficiencies

the challenge

Building a Data Analytics Platform to Scale with LifeScan’s Requirements

One of Lifescan’s most significant challenges was obtaining long-term clinical data accuracy and consistent system performance for blood glucose monitoring of diabetic patients.

A complex & specialized technical challenge

It was critical that the company met and sustained the minimum requirements of the International Organization of Standardization (ISO) and the European harmonized version, demonstrating both product accuracy and consistency.

It was also subject to the Health Insurance Portability and Accountability Act (HIPAA), a regulation that mandates security and privacy for Personal Health Information (PHI). Ensuring HIPAA compliance in a SaaS context was a complex and specialized area, encompassing a variety of technical challenges covering intrusion detection, encryption of data, sophisticated auditing capacity, and OS-level security establishments.

Key Business Requirements:

Perform massive data processing of glucose meter data to generate various reporting metrics and KPIs and gain real-time insights into patients’ health conditions; % of patients with hypoglycemia, hyperglycemia, etc.

Key Technical Requirements:

Cost-effective serverless cloud setup to process glucose meter data up to several GB in size

Design of standardized coding practices for complex ETL logic blocks

Ability to run concurrent queries

Security design

Data retention policies

Provisioning for auditing needs

Pay-per-query cost model to support query load of hundreds of reporting users

Ability to generate analytics with 100% data accuracy, shorter SLAs and cross-domain reporting capability

Enable Industry’s best optimizer, parallel query processing, efficient indexes and several intelligent scan techniques to eliminate the limitation on data accessibility as network of patients and physician grows exponentially in the near future

LifeScan’s end goal was to build and deploy vetted, reusable coding procedures that would save significant time and money while meeting mission-critical objectives.

With the right approach, the company could reduce the cost of on-boarding new diabetic patients by more than half compared to its existing approach.

The Solution

Amazon Web Services (AWS) and Infostretch

LifeScan needed an ETL tool that could process the glucose monitoring datasets with 100% accuracy, enable in-memory iteration on complex data models at the field level, provide serverless interactive query services, and provide flexibility around query modeling.

Infostretch conducted a two-week assessment of detailed business requirement to understand:

Key business-critical functions and how each of them contributed to the overall growth of LifeScan taking into account different service offering, patient interaction mechanisms and feedback channels.
How to best process LifeScan’s large dataset while handling a variety of complex data types and assuring accurate analytics.
The right cloud services to meet LifeScan’s cost, performance, security, operational efficiency, and resiliency requirements.

The right choice

AWS Glue was the right choice because of its ability to provide a fully managed ETL tool employing Spark in-memory engine and serverless analytics platform with a dedicated data frame to handle field-level logics.

Its underlying in-memory engine was fully managed Apache Spark, a distributed processing system primarily utilized for big data workloads. Spark enabled in-memory caching and optimized execution for fast performance. It supported general batch processing, streaming analytics, machine learning, graph databases, and ad hoc queries as well as a pay-per-query to save costs for the growing number of reporting users.

Amazon S3 simple storage was used to save on storage costs, and the Amazon Glue crawler populated the data stores of Amazon DynamoDB through its native interface. It also crawled Amazon API Gateway to create, publish, maintain, monitor, and secure APIs.

The engagement also included building a data lake in which Amazon Athena and AWS Glue were employed to meet various corporate reporting needs with high accuracy. By moving to a standard configuration that could be self-provisioned, the team was able to deploy new code changes, enhancements, and updates to the transformation logic. This shift allowed for the in-time data provisioning needed to bring visibility to some of the key business KPIs and scalable systems to support sudden data growth.

Our Approach

As a part of this project, Infostretch also created compliant infrastructures, error-proof data flow mechanisms, data reconciliation procedures, and reports using an AWS Athena reporting tool.

Testing strategy, planning, and methodology were also critical to ensuring a high level of accuracy in the reports. AWS Enterprise Support also ensured the client’s ability to gain long term access to the product developers behind the services.

Infostretch was able to provide solutions to each of these challenges and enable the company to transform its data analytics and intelligence platform.