ChatGPT and FHIR: The Possibilities

*"HL7, FHIR, and the FHIR [FLAME DESIGN] are the registered trademarks of Health Level Seven International, and their use does not constitute an endorsement by HL7."*

As the world continues to move towards a more connected and digitized healthcare system, the need for standardized data exchange becomes more important than ever before. Healthcare providers, researchers, and patients alike seek more efficient ways to utilize medical data. HL7 FHIR has helped accelerate these healthcare digitization initiatives in recent years. In this blog, we will explore how ChatGPT and FHIR can work together to potentially revolutionize the way healthcare data is utilized to enhance patient outcomes and overall healthcare efficiency.

Why FHIR?

In one of our previous blogs, we discussed the advantages of FHIR and how it is better than many of the current Healthcare document-based data standards like C-CDA. In summary, C-CDA is a structured data document standard containing specific patient data for a particular purpose at a particular time, such as the transition of care. Its limitations lie in the varying data elements and constraints on long-term usability, hindering true healthcare interoperability. FHIR, on the other hand, focuses on granular data called "Resources" that can be easily linked through "references." This decouples data exchange from context, allowing for greater extensibility and adaptability in healthcare data management. This decoupling of data exchange from context is the most significant differentiating factor of FHIR.

The Proof is in the Pudding!

At the start of the year, we went live with over 2000 FHIR APIs to help our clients leverage the benefits of FHIR and empower patients while meeting Cures Act requirements. As the adoption of those APIs continues to grow, we have been getting a lot of questions from our EHR partners and providers about how (and when) they can see the benefits of this “decoupling.” The potential benefits of this decoupling can only be realized when the data is used outside the EHR through SMART on FHIR Apps. Until recently, that meant waiting for the FHIR App ecosystem to grow. Thanks to ChatGPT, we don’t have to anymore!

We are excited to share today our upcoming new solution, “MeldRx Chat,” - a SMART on FHIR app powered by ChatGPT. Check out this overview video of how the power of the modular FHIR Resources can be realized with ChatGPT.

"HL7, FHIR, and the FHIR [FLAME DESIGN] are the registered trademarks of Health Level Seven International, and their use does not constitute an endorsement by HL7."

ChatGPT + FHIR: The Perfect Synergy

Let's consider an example of how ChatGPT and FHIR can transform patient care. In this accompanying video, we take you through Alice’s care journey, who is reviewing her FHIR health records using MeldRx. She wants to learn more about her condition. and asks, "Explain this condition to me." Without requiring additional information, the MeldRx Chat powered by ChatGPT provides Alice with a context-aware response about her renal transplant. But Alice wants to understand her condition in simpler terms, so she asks the MeldRx Chat to explain it in everyday language, and it does. If Alice wants to access this same information in Spanish, the MeldRx Chat can readily provide the translation.

What's more impressive is that the MeldRx Chat can continue to learn and improve as it encounters new data and feedback, enhancing the accuracy and helpfulness of its responses. Suppose Alice wants to know if one of her medications relates to her previously reviewed health condition. She opens the chat within the medication’s context and asks, "Is this medication related to my condition?" The chat remembers her previous conversation and answers accurately without requiring further prompts. This medication wasn’t related to her condition, so she chooses another. This time, MeldRx Chat responds by sharing useful information about how this medication is related to her condition.

Although patient education and empowerment is one possible application, the potential uses of the "MeldRx Chat" app for healthcare providers and researchers are virtually limitless. By utilizing the "EHR Launch," this SMART on FHIR app can seamlessly integrate with any EHR, granting the user access to powerful AI tools for real-time decision support at the point of care with just a single click. Additionally, the app can be used as a “Standalone App” for population health analytics on FHIR repositories like Azure Healthcare Data Services. If you are interested in learning more about our rollout plans with partners, please do not hesitate to contact us.

What About Privacy?

Although privacy is a complex topic when implementing AI in healthcare, FHIR offers a technical solution to help address privacy concerns. Let’s explain how.

While making the calls to ChatGPT in MeldRx Chat, we remove all PHI and substitute it with dummy data. This is simplified due to the structure of FHIR resources. In the demo, we showed chatting about individual data points, aka FHIR Resources like “Condition” or “Medication.” We also allow sharing a complete patient record (as an FHIR bundle) or a filtered list of records. FHIR data can even be mixed with nonstructured data (EHI). ChatGPT can take all that data and even create a patient summary from it. There are already numerous examples of this on the internet. The critical point here is that FHIR helps to control the type of data that is shared and the level of anonymization. We are creating libraries using the FHIR Resource structure that can be extended so a patient, a provider, or a researcher is aware of what level of PHI is being shared.

The Global Reach

In the coming days and weeks, we will continue to see more innovative use of ChatGPT in healthcare worldwide. We firmly believe FHIR will play a significant role in streamlining these initiatives. The Global Digital Health Partnership is an initiative by 33 countries spread across six continents to develop an “International Patient Summary” (IPS) standard that can be used to share patient summaries across international borders. The FHIR community has done a fantastic job of creating an FHIR implementation guide for IPS. This can simplify idea sharing across different apps. Our MeldRx Chat leverages IPS behind the scenes.