Getty Images
ONC Mulls Health Information Exchange Data for Patient-Centered Research
ONC plans to implement data standards, APIs, and machine learning (ML) infrastructure to leverage health information exchange data for patient-centered outcomes research (PCOR) about COVID-19.
ONC has announced a new project to leverage health information exchange (HIE) data to support COVID-19 focused patient-centered outcomes research (PCOR) by implementing new data standards and technology.
State and local HIEs aggregate patient EHR data from more than 60 percent of hospitals in the United States. Yet, these extensive datasets have restricted use because of technical and privacy-related barriers, ONC officials stated.
“HIEs routinely collect patient data from a variety of sources and then facilitate the exchange of patient health information with clinicians, public health agencies, and laboratories,” ONC officials Adam Wong and Wei Chang wrote in the HealthITBuzz blog post.
“Increased use of this data for patient-centered research could help facilitate research activities, including in public health emergencies such as COVID-19,” Wong and Chang continued. “However, varied technical and privacy requirements often put in place by states can make it difficult for HIEs to make data easily usable for researchers.”
The project, “Using Machine Learning Techniques to Enable Health Information Exchange to Support COVID-19-Focused PCOR,” will apply privacy-preserving artificial intelligence (AI) and machine learning techniques to gain access to HIE data to further the understanding of COVID-19.
“The amount of data now available for these types of analyses is greater than ever before, whether from an electronic health record (EHR), a personal device, or environmental sensors,” Wong and Chang said. “While research is just the first step in a process that leads to improved health outcomes, researchers can now use many different types of data sources to inform their work.”
One of the ways ONC plans to leverage HIE data for research is through split learning, a privacy-preserving machine learning technique.
By piloting split learning, ONC will examine the ability to use HIE data for research at the individual HIE level and across multiple HIEs.
ONC hopes to understand if the privacy-preserving machine learning technique is suitable for widespread adoption by HIEs.
ONC will also implement United States Core Data for Interoperability (USCDI) and Bulk FHIR API at three HIEs to facilitate interoperable and efficient data access
HIEs interested in working with ONC will be able to make the most of the Cures Act Final Rule. ONC officials stated that participating HIEs would have access to EHR data from providers using a certified HL7 FHIR API and the USCDI
“The innovative use of the FHIR API standards including HL7 Bulk FHIR API to deliver value in healthcare has never been greater, and we hope that this project lays the ground for further innovation by the research community,” stated Wong and Chang.
The challenge was conducted under ONC's Synthetic Health Data Generation to Accelerate PCOR project supported by HHS' Office of the Secretary PCOR Trust Fund.
Synthetic health data generation tools create artificial datasets that mimic real-world data. Researchers, health IT developers, and informaticians leverage synthetic health data to test new ideas until access to secure and real clinical data are available.
