AWS Security and HIPAA

The Challenges

Cambia Health was faced with two main challenges:

• Unacceptable length of time for their claims processing system to analyze large datasets
• Inefficient management of analytics platforms due to challenges associated with having multiple systems with varied components spread across several different sets of servers, all while maintaining HIPAA compliance

Why Amazon Web Services

Working with Amazon Web Services and 1Strategy, the Cambia Health data science teams have been able to deploy HIPAA-compliant and secure AWS EMR data pipelines. The data pipeline is secured end-to-end with encryption, applied at rest and in transit from the S3 data lake to the EMR cluster and back to an S3 bucket hosting processed data. The ability to utilize these AWS cloud services in a HIPAA-compliant mode for data analysis pipelines has led to increased productivity with ongoing projects and opened the door for Cambia data scientists to explore new friction points within the healthcare domain.

Cambia solicited 1Strategy because of their unique focus on transformation and their depth of knowledge in leveraging the platform services of the AWS ecosystem, as well as their experience with sensitive HIPAA workloads. 1Strategy can quickly distill requirements, provide design expertise and complete automation projects in a fraction of the time estimates provided by their competitors.

The Benefits of AWS and 1Strategy

Speed to Production

The elasticity of the EMR platform allows the data science team to prototype, research and refine analytics products on the same software stack available for full-scale production in a seamless manner. The parity of environments throughout the development cycle has reduced difficult-to-diagnose environmental issues, increased collaboration between the data science and engineering teams and allowed a more wide-spread adoption of best practices.

“Because we were new to the AWS cloud space, we engaged 1Strategy to help jump-start a few initiatives and provide mentoring and knowledge transfer to staff,” said Scott O. Lukas, Cloud Network Architect of Cambia Health Solutions. “Working side by side, we were able to complete our initial IaC (infrastructure as code) implementation of EMR (Elastic MapReduce) and were able to stand up cross account ADFS (Active Directory Federation Service), allowing us to centrally control account access authentication.”

The 1Strategy team helped Cambia design, codify and deploy the services required by Cambia to enable rapid deployment of the AWS services required for the cloud data lake in less than 10 minutes, including EMR clusters and all supporting services.

Lukas added, “I appreciated how eager 1Strategy was to engage with us, openly sharing their knowledge and diving in to help find answers to our questions. I would happily partner with them again in the future.”

Due to the success of this project, 1Strategy and Cambia were invited to share this experience at AWS re:Invent 2016. The session, Data Science and Healthcare: Running Large-Scale Analytics and Machine Learning on AWS (below), was about how 1Strategy was able to work with Cambia to deploy HIPAA-compliant and secure EMR data pipelines on the cloud, as well as utilize machine learning in reducing Cambia’s cost and improving their quality of care. The session was so popular that even the requested encore presentation session was completely full. As a result of this project, 1Strategy was asked to complete three additional automation, DevOps, and data projects for Cambia Health.

The Solutions

Decreased Processing Time for Analytics Pipelines

The runtime for key analytics pipeline processing was reduced from over 24 hours to under two hours. Tens of millions of medical claims records that had previously been analyzed on their legacy, on-premises infrastructure were analyzed on AWS by leveraging modestly sized Amazon EMR clusters (less than 10 nodes of m4.large instances) and built on out-of-the-box EMR images. The EMR toolset allows data scientists to seamlessly apply high-level programming languages to problems throughout the feature engineering pipeline, resulting in more nuanced features that better capture the complexities of the healthcare domain.

Additionally, the decreased time devoted to pipeline management allows the team to dedicate additional iterations toward building and tuning analytics solutions, increasing the leverage of the data science team.

Application Consolidation

EMR provided a flexible processing platform for all data-centric applications in the cloud. Leveraging the high-level languages supported by EMR and the native Hadoop integration, the data science team is able to feed downstream systems reliably without the need for intermediate data formatting and storage.

The ability to customize the contents of the EMR machines has allowed the data science team to move workloads previously spread across multiple servers containing separate software components to a single, orchestrated set of machines for use cases, including clinical prediction, speech-to-text transcription and natural language processing. Combining analytics toolsets such as the Python PyData Stack and R with Big Data Frameworks (primarily Spark) has provided the capability to complete data exploration and feature engineering and applied analysis on a single stack and has vastly improved the efficiency of the team.

The work described in this engagement was originally completed by 1Strategy, a TEKsystems Global Services company acquired in 2019. As of June 2023, 1Strategy has fully integrated with TEKsystems Global Services to continue to deliver AWS expertise to customers. Learn more about our AWS solutions.