G. Cameron Deemer

Using AI to Clean Up the Dirty Little Secret About Healthcare Interoperability

There’s a dirty little secret about interoperability that healthcare leaders don’t often talk about: Despite decades of digitization and standards, we still have a problem with dirty data. That means a great deal of digital information still gets touched by human hands before it can be consumed by hospital or pharmacy systems.

This is more than an inconvenience. Manually entering information can introduce errors that contribute to patient harm and readmissions.

So how can we overcome the need for manual processes that still degrade the sharing of healthcare information between systems? DrFirst is working with leading health systems to implement practical solutions that achieve true semantic interoperability, finally fulfilling the promise of efficient workflows and informed health outcomes. We do it by addressing the four Cs of “perfect data:” complete, clean, consumable, and contextual.

To get complete data, we start with the industry-standard medication history feed, and then work with hospitals to identify where gaps exist for their unique patient population, where their patients are having prescriptions filled at pharmacies that do not contribute to the standard feed. We do the legwork to get those pharmacies on board and sharing data.

To make the data clean and consumable, we apply a patented process of machine learning and artificial intelligence (AI) to properly aggregate duplicate records, safely infer missing data fields, parse free text into structured discrete elements, and map nomenclature so that incoming data matches the vocabulary and databases of the receiving system. The received data becomes semantically interoperable at this point because the hospital EHR can import it properly—without manual intervention—and use it for clinical decision support and analysis.

The final step is to make the data contextual. In other words, the data should support multiple clinical workflows as a result of being truly interoperable. In this step, information is extracted from the data and presented to the user for actionable insights. For DrFirst, this means providing actionable scoring of the data to inform common use cases such as managing opioid therapies or working to improve patient compliance with drug therapy.

This practical, semantic interoperability is helping hospitals and health systems save large amounts of time when gathering and entering medication history while avoiding adverse drug events and medication-related readmissions.

Full interoperability of all healthcare systems remains a far-reaching goal, but advances like these in medication history interoperability are making real, practical improvements today.