Clinicians are drowning in data but starved for actual answers. What if we stopped making them hunt for evidence? The era of standalone clinical reference portals is fading. Bringing verified knowledge directly into the daily clinical workflow is the only path forward.
Healthcare IT Today sat down with Derrick Leung, Business Development Director at BMJ Group, to discuss the challenge of delivering relevant clinical knowledge without adding friction to a provider’s day. The conversation highlighted how the organization is rethinking the delivery of medical evidence.
What This Conversation Revealed
- Integration beats standalone applications. Clinicians do not have time to switch screens to search for medical evidence. Delivering information directly into their existing workflow is now a necessity.
- Human curation acts as an AI guardrail. Large language models are prone to hallucination. Grounding AI tools with deeply-vetted, expert-curated content provides a necessary safety net for clinical decision-making.
- Relationships perform better than rigid rules. Patients present with complex symptoms rather than neat data points. Using a knowledge graph to map these relationships creates adaptable and accurate clinical decision support.
Stop Forcing Clinicians to Search for Clinical Information
The BMJ (formerly British Medical Journal), the renowened peer-reviewed medical journal by BMJ Group has a nearly 200-year history of publishing medical evidence. Now, they are moving that knowledge base directly into the clinical workflow via an API. The goal is to eliminate the friction of forcing doctors to open yet another application.
Leung pointed out the value of this approach, “Instead of a clinician going to a different application to search for evidence, which they can still do, they now have that evidence served up within the workflow. Now they aren’t disrupted from their work. That’s powerful.”
By embedding knowledge where physicians already are, organizations can reduce cognitive load.
Human Curation as an AI Guardrail
Technology companies are racing to deploy large language models in healthcare. Yet these models desperately need reliable anchors to ensure patient safety. BMJ is positioning its massive library of evidence as a foundational layer to keep AI-driven clinical tools accurate and relevant.
Leung sees this human curated content base as an important hedge against hallucinations which are inherent to AI tools. “We’re still doing human curated content so that BMJ Group can act as a guardrail for AI technologies.”
Relationships Over Rigid Rules
Traditional clinical decision support systems are often built on rigid rules. But patients present with messy symptoms, not neat, binary data points.
Recognizing this reality, BMJ Group is using a dynamic knowledge graph. This architecture allows applications to organically map a patient’s real-world presentation to the right clinical evidence.
Leung described the architectural this way: “Our knowledge graph is not like traditional clinical decision support which is rules-based. Our knowledge graph is based on relationships.”
Knowledge graphs are a structured, graph-based representation of entities and their relationships. This makes them stronger at storing complex, connected facts whereas rule-based systems are better at arriving at a defined answer based on if-then logic. By using knowledge graphs, BMJ Group allows for a more flexible and realistic approach to diagnostic support.
The Bottom Line
Delivering clinical evidence is no longer just about publishing accurate information. It is about making that evidence useful. It needs to be in the right format, presented in the right context, and be as easily accessible as possible (aka zero disruption to clinicians). By embedding knowledge directly into workflows, grounding AI with human curation, and mapping relationships instead of rules, BMJ Group is positioning itself as the go-to partner for clinical evidence in the age of AI in healthcare.
What Healthcare IT Leaders Are Asking
How can health systems deliver clinical evidence without disrupting workflows?
Rather than forcing clinicians to log into separate reference portals, organizations are integrating evidence directly into the electronic health record via APIs. This embedded approach surfaces relevant medical data precisely when the clinician needs it, reducing cognitive load and saving valuable time.
What is the best way to prevent hallucinations in clinical AI tools?
AI models must be anchored to verified, human-curated medical content. By using an established, peer-reviewed knowledge base as a strict guardrail, technology vendors can prevent their tools from generating unsafe or inaccurate clinical recommendations.
Why are knowledge graphs replacing rule-based clinical decision support?
Rule-based systems rely on strict if-then logic, which struggles with the complex, overlapping symptoms of real patients. Knowledge graphs map the dynamic relationships between symptoms, diseases, and treatments to provide a more flexible and accurate diagnostic pathway.
Learn more about BMJ Group at https://bmjgroup.com/
Listen and subscribe to the Healthcare IT Today Interviews Podcast to hear all the latest insights from experts in healthcare IT.
And for an exclusive look at our top stories, subscribe to our newsletter and YouTube.
Tell us what you think. Contact us here or on Twitter at @hcitoday. And if you’re interested in advertising with us, check out our various advertising packages and request our Media Kit.
No comments:
Post a Comment