Skip links

On-Demand Webinar: To Learn How Payers Can Automate Prior Auth at Scale with AI. View Now.

Industry blog

How Improved Interoperability Can Help Healthcare AI Flourish

As patients, we’ve all felt the pain of trying to transfer medical records. Between the HIPAA release forms, wait times, forgotten requests, and antiquated transfer methods (faxes and CDs are still common methods in healthcare), the process of trying to give a new provider our health data is filled with friction and blindspots. This friction exists between organizations inside the healthcare system as well. Insurance plans struggle to obtain clinical data from providers, providers struggle to access plans’ more extensive member claims data, and healthcare vendors struggle to access data from both sides, even with the appropriate agreements in place.

With CMS’s big push around electronic health record (EHR) adoption in 2010 as part of its Meaningful Use rule, everyone thought the old, manual ways of sharing healthcare data would disappear. However, because Meaningful Use didn’t provide incentives around data sharing or clear interoperability standards, these new EHR systems didn’t help data flow more freely between healthcare entities—in many ways, they made things worse. Today, most provider groups have their own EHR systems that makes data sharing and extraction challenging. The same is largely true of health insurance claims processing and administrative systems. The sad truth is that, while plans and providers are generating a ton of electronic healthcare data, most of it isn’t easily accessible or usable—even within the walls of a single organization. In fact, as much of 80% of your healthcare data goes unused in the provision of patient care.

In recent years, the industry has sustained a focused conversation around interoperability—a term describing a connected healthcare system in which data is exchanged freely and made easily accessible to patients, plans, providers, medical device manufacturers, and vendors. However, in spite of widespread interest around interoperability, there hasn’t been much progress toward making it a reality. And today, in the brave new world of IT advances such as artificial intelligence (AI), the lack of interoperability has become a massive roadblock for tech-powered improvements to clinical care, administrative efficiency, and patient health management. 

How Did We Get Here?

On the surface, interoperability sounds like a goal every healthcare organization should share. Who wouldn’t want easier access to more data to improve patient care and business operations? But, as with many things in healthcare, what seems simple on the surface is more complicated in reality. 

For better or worse, the history of healthcare data is one of segregation. Before cloud-based systems gained widespread adoption, healthcare information was stored in on-premises, siloed repositories that could only be accessed by people within the organization. This reinforced the behavior that patient or member data housed by a healthcare entity is “owned” by that entity—and that the “owner” therefore controls who can access that data. Additionally, given the ambiguous HIPAA guidelines around patient privacy and security, healthcare organizations have been even less inclined to share data with each other, even when this information is needed to administer care.

EHRs were supposed to change this siloed approach to data storage and commoditization of patient data. Unfortunately, market dynamics have been such that EHR companies benefit from keeping data locked up inside their own systems, so many platforms have made it difficult to share or export patient information. This means we still have data silos—but instead of physical filing cabinets or servers, we have localized digital systems that don’t easily integrate with other platforms.

The Regulatory Push for Interoperability

Over the past few years, CMS and HHS have seen the distance grow between their vision of a connected healthcare data ecosystem and reality. A variety of proposals have been made to promote interoperability, with varying levels of technical specificity around implementation and potential consequences for perpetuating the status quo. In 2019, two government healthcare agencies proposed regulations that are putting increasing pressure on healthcare technology vendors, payers, and providers to move toward interoperability.

CMS’s Interoperability and Patient Access Proposed Rule

This proposed rule, put forth by the Center for Medicare and Medicaid Services (CMS) in February 2019, promotes data sharing between healthcare entities and with patients. There are a few different components to the proposed rule: 

  • MyEHealthData initiative / Blue Button 2.0 API: Makes patient claims data available via consumer apps using HL7/FHIR. Implementation is already happening for Medicare fee-for-service claims; this rule would extend the work to Medicare Advantage, Medicaid, and CHIP data in the future.
  • Health Information Exchange and Care Coordination Across Payers: Requires participation in a data exchange network for payers across all government plans (Medicare/MA/Medicaid/CHIP/QHPs).
  • API Access to Published Provider Directory Data: Requires all Medicare, MA, CHIP, and Medicaid payers to share their provider directory data publicly via an API. 
  • Trusted Exchange Network: Requires all payers in CMS programs to share data through a trusted exchange network so information can flow securely and privately between plans and providers throughout the healthcare system.

ONC’s 21st Century Cures Act: Interoperability, Information Blocking, and the ONC Health IT Certification Program 

This proposed rule, put forth by the Office of the National Coordinator for Health Information Technology (ONC), implements certain aspects of the 21st Century Cures Act and introduces new requirements to promote seamless and secure access, exchange, and use of electronic health information (EHI). A few of the key provisions of this proposed rule include:

  • Information Blocking: Mandates secure data sharing between covered entities (payers, providers, and vendors) for all reasonable and necessary healthcare activities. Organizations who continue to block information will be subject to $1 million fines or more. 
  • Adoption of the United States Core Data for Interoperability (USCDI) Standard: Establishes a set of data classes and constituent data elements that would be required to be exchanged in support of interoperability nationwide.
  • EHI Export: Enables the export of EHI for a single patient upon request and supports the export of EHI when a health care provider chooses to transition or migrate information to another health IT system.
  • APIs: New HL7 and FHIR API requirements are designed to support services that give access to data for individual patients and multiple patients.

While payers and providers have been receptive to both interoperability initiatives, there’s still debate around implementation timelines and requirements. However, whether healthcare organizations begin rolling out these changes in 2020 or some later date, CMS and HHS are clearly committed to increasing access to healthcare data.

Interoperability Benefits for Artificial Intelligence (AI)

It’s clear that interoperability is a priority for government healthcare agencies. But what impact will these forthcoming rules have on IT vendors and internal tech departments that want to reap the benefits of AI in healthcare? In short: Increased interoperability will benefit any organization using AI techniques to analyze data at scale. 

Data blocking restrictions will open the way for more data exploration and give access to large datasets that AI algorithms require for training, learning, and adaptation. New APIs and exchange networks will make it easier for approved healthcare entities to access data from other organizations within the industry, allowing meaningful information to flow more easily throughout the system. In the future, some of the infrastructure being built to support interoperability could also provide a platform for sharing AI-driven insights between organizations. Standardized data classes and elements will simplify the process of cleaning, labeling, and running models on existing healthcare datasets, allowing AI models to be deployed faster. All of these forthcoming changes will make it easier for technology vendors and plan/provider IT departments to extract value from healthcare data. 

Paired with regulatory changes around data use, governance, privacy, and security to ensure organizations can use all this newly-accessible data in appropriate ways, interoperability improvements will facilitate greater adoption of AI solutions and improve the accuracy and value of AI applications for the healthcare industry. The next few years of interoperability improvements will be pivotal in paving the way for AI advancement and adoption.

Apixio uses OCR, NLP, machine learning, deep learning, and other AI techniques to extract insights from unstructured healthcare data. Learn more about our technology platform.

🍪 This website uses cookies to improve your web experience.