AWS Government, Education, & Nonprofits Blog

Patient-centered health: How AWS helps patients take control of their own health data

By Dr. Taha A. Kass-Hout, Arun Ravi, Melanie Kaplan, and Pat Combes

Imagine: Transforming the patient and caregiver experience by streamlining health providers and administrative staff interactions and calculating risk score and clinical decision support systems to offer clinicians actionable insights at the point of care. Establishing healthcare data interoperability interfaces, like Fast Healthcare Interoperability Resource (FHIR), can help empower patients and improve their care.

Backed by the nonprofit HL7, the FHIR application programming interface (API) facilitates data exchange between enterprises, like health systems, and uses medical claims for analytics. FHIR helps software developers build applications to benefit patients and clinicians, like a secure application for patients to pull data into a portal of choice. This approach reduces friction, introduces automation, and provides new methods for delivering cost-effective services to close gaps in care. Patients become active, informed partners with their clinicians.

Last year, Amazon Web Services (AWS) worked with Fred Hutchinson to create a FHIR-enabled storage and APIs, enabling care coordination between oncologists and primary care providers. Fred Hutch used the APIs to provide patients with an application to support their regimes, including appointment follow-up and engagements with multiple providers, providing visibility into disease progression. This digital-therapeutic approach helped improve patients’ mental health, health outcomes, and overall experience.

Technical integration: Achieving syntactic interoperability

Most electronic health record systems (EHRs) do not follow patients on their journey of care beyond the hospital walls. As a result, a patchwork of healthcare data emerges. The average health system in the United States (US) struggles with integrating data and coordinating care across as many as 18 different EHR systems among its various affiliated providers. FHIR can help integrate the fragmented pieces of patients’ records. FHIR combines the features of existing standards specification – such as HL7 V2, HL7 V3, and CDA – while using the internet to exchange information and democratize the flow of information. Based on RESTful web services, the FHIR design uses multiple standards like HTTP, JSON, URL, or XML in contrast to the majority of IHE profiles, which rely on SOAP web services protocol (largely based on HTML or XML). FHIR also migrates data to various messaging standards, such as from HL7 v2 messages and CDA documents.

In the US, the Office of the National Coordinator for Health Information Technology (ONC) and the Centers for Medicare and Medicaid Services (CMS) champion an open standard, generated by the user community to achieve greater interoperability and rapid exchange of data providing patients “safe, secure access to, and control over, their healthcare data.” CMS advocates for Blue Button 2.0 adoption, with FHIR at its core, to benefit patients and their care teams (Mirth/Nextgen Connect for bi-directional exchange of messages). Consented information flows into the NIH All of Us Program to realize personalized medicine at the individual level.

Internationally, the United Kingdom National Health Service (UK NHS) selected FHIR as the standard for exchanging information, and the Nordic Council of Ministers’ eHealth group published a report with guidance for setting up interoperable digital public services with open standards (technical and semantic) such as FHIR and openEHR.

SMART on FHIR

The SMART open specifications offer developers a framework to create, authenticate, and integrate a healthcare application with any organization regardless of the underlying EHR system. For example, to display a patient’s blood pressure measurements from the past six months at home and in the clinic, you can build a provider view inside the EHR and a consumer application, exchanging data using the FHIR data model. FHIR supports data ingests (with patient’s consent) from EHRs, patient generated data or devices (wearables, digital therapeutics), and applications (patient portals, health coaches).

An app could display data trends, highlighting anomalies at the individual or population level. SMART’s inclusion in proposed federal rules will provide ways for patient-facing technology to better integrate with EHRs and for the bi-directional exchange of information. Over the past five years, Redox, an AWS customer, has enabled its healthcare customers to use SMART on FHIR to launch applications inside the EHR while exchanging patient-authorized data with Single Sign On (SSO).

Capturing context: Achieving semantic interoperability

The majority of the data in medical records today consists of unstructured narratives and semantics captured in text, voice, image, PDF, or scan formats. Discharge instructions, radiology reports, or operative notes improve patient trajectory prediction, enhance care team coordination and understanding of social determinants of health, related inequalities, and the vital conditions for health (and wellbeing) that are key to personalize actions and adjust risks for each patient.

AWS developed a solution using Amazon Comprehend Medical to extract medical conditions from bulk medical notes via exchanges using FHIR Bulk Data Access (Flat FHIR) with the DocumentReference resource (currently, FHIR USCDI-based resources do not include notes.) This solution demonstrates ingesting bulk HL7 Medical Document Management (MDM) messages (or Media, which includes free text notes, PDFs, images, scans, etc.) then map data directly to FHIR resources.

We believe future engagements in healthcare will be revamped using free-form, voice interactions with voice-enabled devices like Amazon Alexa and smart devices. Our solution can process various media modalities by transforming the data input from voice, image, scan, PDF, etc. into text. Then, Amazon Comprehend Medical processes the text and maps into clinical FHIR resources surfaced by the API endpoints. For example, Textract can process scanned documents, and Amazon Transcribe Medical can transcribe recorded medical dictations that can then be further structured using Amazon Comprehend Medical.

With access to all available information, advanced analytics and machine learning (ML) can enhance medical and scientific insights tied to patient outcomes in an accurate, scalable, secure, and timely manner. Vancouver General Hospital (VGH) and University of British Columbia (UBC) researchers leverage Amazon Comprehend Medical and Amazon SageMaker to create their own machine learning models for triaging x-rays to provide a better experience for patients and providers. New chest x-rays from the emergency room are automatically processed by the model, which radiologists then triage, or de-identify and share for research and learning purposes. VGH achieves efficiency, cost savings, and high accuracy at scale by applying machine learning combined with physician’s input and observations.

Compliance

Many healthcare and life science customers choose AWS because of our proven record to meet and enable compliance requirements and secure data in highly regulated industries. AWS supports a number of security standards and certifications relevant to healthcare, including HIPAA, HIPAA eligible services, FedRAMP, EU General Data Protection Regulation (GDPR), HITRUST, and ASIP-HDS. No customer data is used to train or improve the machine learning models in Amazon Comprehend Medical.

Conclusion

Patient-centered health means incorporating patients into their own journey of care and empowering patients, clinicians, and caregivers, with the information and tools they need to improve health outcomes. Today, this may look like a patchwork quilt—a health history stitched together with any data that can be converted into a computable format. Tomorrow, the data will flow more naturally where FHIR and supplementary platforms like SMART are key components in helping patients and healthcare providers together to manage, mitigate, and ultimately cure diseases.

Learn more about Amazon Comprehend Medical, Amazon Transcribe Medical, and healthcare and life sciences at AWS.

Taha A. Kass-Hout, MD, MS

Taha A. Kass-Hout, MD, MS

Kass-Hout, MD, MS, is general manager, healthcare and artificial intelligence (AI), and Chief Medical Officer at Amazon, focusing on healthcare and AI-related initiatives, including Amazon Comprehend Medical, Amazon’s first health care-specific machine learning service offered by AWS. Taha received his medical training at Beth Israel Deaconess Medical Center, Harvard Medical School, and during his time there, was part of the BOAT clinical trial. He holds a Doctor of Medicine and Master of Science (Bioinformatics) from the University of Texas Health Science Center at Houston.