Build population health systems to enhance healthcare customer experiences on AWS
As the amount of health data increases, different healthcare, life sciences, population health, and public health organizations are working to modernize their data infrastructure, unify their data, and innovate faster with technologies like artificial intelligence and machine learning (AI/ML). In this blog post, we dive deep on architecture guidance that enables healthcare providers to improve patient care. Learn how Amazon Web Services (AWS) supports both population health and public health organizations to improve customer experience (CX) and Service Delivery across multiple domains. The latter include medical care coverage, public health interventions, genetics-based therapy, improving home monitoring for chronic diseases, using patient-generated health data to fill in Electronic Health Records (EHR) gaps, and predictive analytics based on social and physical environments like urban design, clean air, and water.
Reduce the paperwork burden on citizens
Needs vary across partners within healthcare, population health, and public health. But a common denominator is that they benefit and provide better CX to those they serve, including veterans, elderly seeking Medicare, and Medicaid services recipients, when systems are interoperable and leverage analytics and ML. For example, instead of trying to stitch together disparate medical records from each of their doctors, polytrauma patients can benefit from integrated history and analytics across all care providers for continuity of care and improved diagnostics. Population health practitioners are challenged with pulling more and more varied data such as EHR, lab reports, monitoring device data, and more, to drive better population health outcomes. Add to that the complexity associated with social determinants of health (SDOH) and environmental issues like air quality data, that impact the health of individuals or large groups.
Build the architecture pattern
In order to support the preceding use cases, we need a broad range of capabilities working in conjunction, as illustrated in Figure 1, from data ingestion to normalization, storage, and finally data visualization.
As a foundation, and to meet Office of the National Coordinator for Health Information Technology (ONC) and Centers for Medicare & Medicaid Services (CMS) interoperability and patient access rules, build the central data lake using AWS HealthLake. This is a HIPAA-eligible service that supports the shifts towards using more Fast Healthcare Interoperability Resources (FHIR) formatted information while integrating seamlessly with analytics and machine learning solutions. This setup allows organizations to easily share with each other using a common format. Use Amazon Simple Storage Service (Amazon S3) for unstructured data that doesn’t naturally fit into the FHIR format or structured publicly available datasets like those available through healthdata.gov or AWS Data Exchange.
Dive deep on architecture sub-components
With AWS HealthLake, you can interact with the data at both patient and population level, depending on your use case. The service supports a number of FHIR-based APIs. These APIs allow you to create, read, update, and delete (CRUD) individual or multiple records including building integrations with your data. You can access APIs via Amazon API Gateway, which integrates with AWS Lambda for authentication. Alternatively, you can access the APIs directly inside of your account, if you are building your application directly on top of AWS HealthLake.
For bulk FHIR data ingestion, there are a number of options (see Figure 2), which depend on your use case. In either case, you would do the initial move into Amazon S3, for instance by using Amazon Kinesis for streaming data or AWS DataSync for bulk data. Refer to this whitepaper for a deeper dive on how to move data to Amazon S3 with the above or additional AWS services. In either case, data is encrypted in transit and at rest using AWS Key Management Service (AWS KMS).
AWS HealthLake requires data to be formatted in the FHIR format. If data is ingested in other formats like Consolidated Clinical Document Architecture (C-CDA), data can be converted as a part of a data pipeline (using either AWS Partner solutions through the AWS Marketplace or building your own through services like AWS Lambda).
Data normalization, information extraction, and storage
Store FHIR data in AWS HealthLake and non-FHIR data like large documents or public data sets in Amazon S3.
AWS HealthLake can automatically extract medical information from unstructured medical text like doctors’ notes, clinical trial reports, or radiology reports (see Figure 3) as well as identify relationships among extracted health information and link to medical ontologies like ICD-10-CM, RxNorm, and SNOMED CT. This all happens through its integration with Amazon Comprehend Medical, which helps to significantly reduce time and effort of healthcare professionals, allowing them to focus on higher value tasks and improving the customer experience.
If you want to extract text from documents like health intake forms and insurance claims prior to loading them into AWS HealthLake or for storing large documents separately, you can use Amazon Textract, which uses AI to extract text from various document formats.
AWS Glue is a serverless data integration service that makes it easier to discover, prepare, move, and integrate data from multiple sources for downstream analytics and ML at any scale.
Finally, access to data for analytics and ML purposes is governed with AWS Lake Formation. AWS Lake Formation provides data mesh capabilities to secure, manage, and audit access to the data within or across AWS accounts.
Data analytics and visualization
When you create an AWS HealthLake data store, the highly nested FHIR data structure is ingested into Amazon Athena and automatically transformed into Iceberg tables queryable with SQL (see Figure 4). Since Amazon Athena integrates with AWS Lake Formation and AWS Glue, this makes it easy to access and transform data for analytics and ML purposes.
AWS offers the broadest selection of analytics services that fit all your data analytics needs including big data analytics, machine learning, and more. This includes Amazon SageMaker to build, train, and deploy machine learning models for any use case with fully managed infrastructure, tools, and workflows. Or, you can use Amazon SageMaker Canvas to generate accurate ML predictions with no code at all.
The Shared Responsibility Model provides an overview of how security and compliance responsibilities are shared between AWS and the customer. Use the design principles of the Security Pillar in the AWS Well-Architected Framework to build a security framework that can protect valuable data and dive deeper with the AWS Security Reference Architecture for security best practices and guidelines in a multi-account environment.
AWS can help healthcare providers, population health, and public health organizations deliver for customers
With this scalable architecture guidance, US federal agencies can improve CX for an individual patient as well as at the population level. This applies whether the customer is a patient needing better insights into their own data across providers, or a healthcare practitioner working to improve healthcare outcomes across a larger population. AWS HealthLake and supporting analytics and ML services enhance customer experience by improving interoperability, and by extracting and analyzing important information from clinical notes, lab reports, and more —improving efficiency and effectiveness for physicians and others working on the data to improve customer care and population health improvement initiatives.
Contact your account team or reach out to the AWS public sector team to learn more about how to use the above reference architecture for a proof-of-concept that fits to your customers and your organization’s needs.
For additional information on improving CX, read about how the cloud enables transformational citizen experiences, and dive deep on an architecture framework for transforming federal customer experience and service delivery, how to improve government customer experience by building a modern serverless web application in AWS GovCloud (US), and architecture patterns for high-traffic public services.