The Semantic Gap.
Why your data scientists keep saying “we have to understand this data first.”
|
|
By the Axiomera team
·
April 30, 2026
·
7 min read
|
|
|
|
The Argument
Healthcare data has a quiet problem.
For fifteen years, the industry has invested in moving data faster. Pipelines, lakes, warehouses, real-time streaming, query engines that answer billion-row questions in seconds. Speed and scale are solved. The industry invested — correctly — in movement.
And we still have a meaning problem.
The same patient concept is represented differently across EHRs, billing systems, lab platforms, and payer feeds. Standards are partially applied, inconsistently interpreted, applied too late. Every analytics project starts with the same sentence: “we have to understand this data first.”
You can’t fix that with another pipeline. You fix it with a semantic layer that runs upstream of every consumer.
|
|
Same procedure · Three source systems
|
The semantic gap
Same procedure. Three representations. No shared meaning.
|
|
Downstream
Every analytics project starts here: “we have to understand this data first.”
|
|
|
Data volume has scaled beyond data meaning.
|
|
|
|
In the Field
What this looks like at scale.
A national real-world data aggregator runs the largest hospital-based RWD asset outside government. Their data estate has the same quiet problem at industrial scale.
|
1,400+
contributing
hospitals
|
358M
unique
patients
|
20yr
longitudinal
window
|
1
shared
meaning?
|
Each hospital maintains its own Chargemaster. Every longitudinal cohort crosses the October 2015 ICD-9 to ICD-10 boundary. Pharma customers are asking for FDA-grade lineage on every record.
The industry’s answer for a decade has been “migrate it to the lakehouse.” But the lakehouse amplifies speed, not meaning.
The work happens upstream — bind every field to an ontology, preserve confidence and evidence, harmonize across sources and across time, and deliver it to the destination with provenance attached.
|
|
|
Why now
The regulatory clock is running.
FDA’s Real-World Evidence Program is no longer theoretical. Since the 21st Century Cures Act, regulatory submissions using RWE have steadily accelerated. Every submission needs auditable, versioned, evidence-backed lineage for every semantic decision the data went through to reach the analysis.
Most data platforms were not designed with that requirement in mind. Retrofitting it is expensive. The infrastructure pressure is becoming a near-term operational pressure.
| CMS-0057-F |
Prior Authorization API. FHIR-native prior auth APIs required by Jan 2027. |
|
| CMS-9115-F |
Interoperability and Patient Access. Patient data exchange via FHIR APIs across payers and providers. |
|
| FDA RWE |
Real-World Evidence Program. 21st Century Cures Act framework expanded by FDA guidance and JAMA 2018 (Corrigan-Curay et al.). |
|
|
|
|
|
|
|
The science
The platform you buy should be reducible to published science.
In semantic intelligence, there is a lot of language about “intelligent data” and “AI-ready pipelines” — almost none of it is backed by published, peer-reviewed work. Because most of it isn’t backed by work.
Axiomera rests on Data Field Theory, an original mathematical framework developed by Dr. Reza Nehzati, Ph.D., our AI Research Lead. Eleven peer-reviewed papers and eight patents pending across US, European, and PCT jurisdictions. The math is open. Your enterprise security teams can audit it.
|
From the paper portfolio
|
|
Semantic Classification Engine
Ontology binding with calibrated confidence scoring
|
|
Data Harmonization Model
Harmonization across sources and time, including coding-system transitions
|
|
Quantum-Superposition Clinical Intelligence Networks
Federated harmonization across institutions, with federated learning
|
|
Read the full research portfolio →
|
|
|
|
|
Issue 02 — The Semantic Gap white paper.
The full argument, in twelve pages. In your inbox the day it lands.
|
Axiomera Research
The
Semantic
Gap.
White Paper
12 pp · 2026
|
|
Inside the paper
| 01 |
The semantic gap, defined. |
|
| 02 |
Healthcare data at scale — RWD, pharma, payers. |
|
|
|
| 04 |
Data Field Theory — an open mathematical framework. |
|
|
|
|
|
|
Want to talk through this?
Thirty minutes with the Axiomera team to talk through whether the platform fits the way your organization runs data.
Or just reply to this email if anything sparked a question. It reaches the team directly.
/ The Axiomera team
|
