The following is a guest article by Dr. Kaushal Kulkarni, M.D., Physician and Chief Medical Officer at Predoc
In the computer science world, there is a commonly discussed problem known as the semantic gap. It’s the idea that the way computers process information is radically different from the way a human understands information.
Take an image, for example. To a computer, a photo of a dog is a series of pixels and patterns. A person, on the other hand, not only sees the image, they project memories, emotions, senses, cultural questions and more onto an instantly recognizable image. See the difference?
In healthcare, the semantic gap isn’t just a philosophical mismatch between data and meaning. It’s operational. It shows up every single day in the gap between having data and being able to use it.
On paper, the industry has made enormous progress in creating and moving patient data, particularly in recent years with the 21st Century Cures Act and the more recent CMS Health Tech Ecosystem initiatives. And yet, when a patient is sitting in front of a clinician today, the data is just not there in a usable form. So humans step in. Then they read, interpret, and manually reconstruct the patient’s story so care can move forward. When the full story is not there, they make phone calls, request faxes, chase records, and open PDFs.
That’s the semantic gap in healthcare.
It’s the difference between accessible data and usable information. And today, it is being filled not by technology, but by people.
The Semantic Gap
APIs, the rise of FHIR as a standard for exchange, national networks, and regulatory pressure have all pushed the industry toward greater interoperability. And if data can move, the problem is solved, right?
Except for the fact that healthcare data is still not in any one standardized format. Behind the scenes is an unfathomable amount of information (now measured not in terabytes or even exabytes, but zettabytes) living in faxed documents, scanned PDFs, unstructured notes, and disconnected systems and portals. Even in an increasingly digital ecosystem, critical pieces of the patient story aren’t in a format that is easily searchable and sometimes not really accessible with the plug and play tools available.
If, for example, a patient’s record is in the form of hundreds of physical pages, clinicians are left to track it down, have it faxed over, and sort through page after page to piece together what the data actually matters. A single data point buried deep in the stack could change everything. But getting to that point can take hours, while other patients wait and care is delayed.
What Gets Stuck
When this semantic gap stays open, the effects ripple across the entire healthcare system.
Clinician burnout, operational inefficiency, delays in care, repeat testing, and poor patient experience all trace back to the same root cause. We’re spending hours tracking down records, reviewing fragmented information, and piecing together an incomplete picture, all before a doctor can even get started on delivering care. Entire teams are built around retrieval and intake just to make the system function. What should take seconds can take days, and when information is missing, the system compensates with duplication and guesswork – in the end, it is the patients who suffer.
The semantic gap explains why so many AI initiatives struggle to move beyond pilots. These systems depend on complete, structured, and trustworthy data. Without that foundation, even the most advanced models fail in real workflows.
There are clear examples of what happens when this is addressed. Organizations that have improved how data is retrieved and prepared have reduced intake times from days to hours and increased patient capacity without adding clinical staff.
Closing the Gap
The healthcare industry is not short on data. It’s not even short on ways to move data.
What it needs is a consistent way to ensure that when data arrives, it is complete, structured, and ready to be acted upon. Until that changes, every gain in interoperability will continue to fall short of its promise.
There will always be administrative work in healthcare. But the current model, where people are required to retrieve, interpret, and reconstruct the patient story before care can begin, is not sustainable. It does not scale with growing data volumes, rising patient demand, or the expectations being placed on the system.
Right now, progress depends on human effort filling in the gaps left by technology. That slows care, limits access, and places a ceiling on what the system can achieve.
The trajectory is clear. More data, more complexity, and more pressure to do more with less. Without a fundamental shift, the gap only widens.
We can’t build the future of healthcare on data that isn’t usable.
Closing the semantic gap is the prerequisite for everything that comes next.
Dr. Kaushal Kulkarni is Co-Founder and Chief Medical Officer at Predoc, where he focuses on making healthcare data more complete, usable, and accessible for clinicians. A board-certified ophthalmologist with subspecialty training in neuro-ophthalmology, he has practiced across academic and community settings and brings firsthand clinical perspective to interoperability and health data workflow challenges. He earned his M.D. from Rutgers Robert Wood Johnson Medical School, completed residency at Georgetown University, and finished fellowship training at the University of Miami’s Bascom Palmer Eye Institute. He is a vocal advocate for usable patient data and clinician-centered innovation.
No comments:
Post a Comment