Communicating a patient's diagnosis through modern tools like patient portals isn't always straightforward. Consider a broken foot, for example. While most patients may expect a simple confirmation, the full radiology report must describe the fracture in precise medical terms -- identifying which bone is broken, the fracture type, whether it's displaced, and if it involves a joint.
Jordan Bazinsky, CEO of Intelerad
These technical details are essential for providers treating the injury, but they often overwhelm patients. Ideally, a clinician would deliver the report along with a clear explanation. Yet patients waiting days for routine screening results or hoping to share their images with loved ones may find themselves without a "radiology translator."
To help bridge this gap, Stanford University developed RadGPT, a large language model (LLM) that generates concept-based explanations to make imaging findings more understandable for patients. This breakthrough highlights a blind spot in healthcare technology. The imaging industry has largely focused on using AI to streamline clinician workflows and support decision-making, but the patient-facing side of the technology has not been adequately explored. This limitation highlights an opportunity to improve health literacy, patient engagement, and trust, and calls for a reevaluation of priorities across the industry.
Comprehension challenges
The 21st Century Cures Act gave patients access to their radiology reports, and while well-intentioned, this mandate for transparency can create confusion rather than empowerment. Patients who don't understand their results may struggle to follow care plans, experience unnecessary anxiety, and be unable to participate meaningfully in treatment decisions. With RadGPT's translation capabilities now demonstrating what is possible, radiology leaders have an opportunity to reimagine patient engagement through AI-powered communication tools. The technology could be used to do the following:
- Personalize educational content to match patients' literacy levels and preferences
- Enhance patient portals to anticipate questions and provide information
- Generate predictive engagement models to identify patients at risk of non-adherence and proactively address concerns
The goal is to enhance, not replace, communication between radiologists and patients. When patients arrive at appointments with a basic understanding of their results, radiologists can devote more time to offering analysis, planning treatments, and building relationships.
Ripe for communication transformation
The potential of patient-facing AI extends far beyond medical imaging. Complex results in pathology reports, cardiology studies, orthopedic evaluations, and other specialties all challenge patient comprehension. Departments that solve this problem first stand to improve patient satisfaction, engagement, and outcomes.
There are no standard protocols or best practices for implementing a program that leverages patient-facing AI tools. Successfully launching an initiative requires balancing innovation and strict attention to safety, accuracy, and regulatory compliance. The technology must be rigorously tested, monitored, and designed with built-in safeguards to prevent misinterpretation.
Adopting a holistic approach that positions AI to serve the entire care ecosystem won't be easy. Ultimately, solutions that advance both provider efficiency and patient confidence are essential-- not just for medical imaging, but as a model for the future of healthcare.
Jordan Bazinsky is CEO of Intelerad.
The comments and observations expressed herein do not necessarily reflect the opinions of AuntMinnie.com, nor should they be construed as an endorsement or admonishment of any particular vendor, analyst, industry consultant, or consulting group.
