Zero-copy health data architecture: how trust does not require data movement
Most health data architectures assume that sharing data means moving it. Zero-copy architecture breaks that assumption by allowing queries to execute where data already lives, eliminating duplication and the trust violations that come with it. But zero-copy alone does not solve the trust problem unless every record queried carries a verifiable trust score.
Moving health data between systems is the default assumption in healthcare IT. Every integration, every analytics project, every AI training pipeline begins with extraction, transformation, and loading into a new location. That assumption is the root cause of most trust failures in health data today.
Zero-copy health data architecture offers a different model. Instead of replicating records across environments, it allows compute to travel to the data. The records stay where they originated. The query executes in place. No duplication. No drift. No orphaned copies living in staging tables that nobody audits.
But zero-copy alone is incomplete. A record that never moves can still be untrustworthy. The architecture solves the movement problem. It does not solve the trust problem.
How does zero copy data work?
Zero-copy architecture enables one system to query another system's data store directly, without creating a physical copy of the underlying records. The requesting system sends a query. The originating system executes it and returns results. The data itself never leaves its source environment.
This pattern has gained traction in enterprise platforms. Salesforce Data Cloud Zero Copy Connectors, for example, allow analytics tools to query data lakes and warehouses without importing records into Salesforce. Snowflake, Databricks, and BigQuery all support similar patterns through data sharing protocols.
In healthcare, zero-copy matters because every copy of a patient record is a liability. Each replica introduces drift, creates a new attack surface, and fragments the chain of custody. A single patient record duplicated across five systems becomes five records that can diverge independently. Zero-copy eliminates that divergence at the infrastructure level.
The health data architecture problem zero-copy partially solves
Traditional health data architecture relies on four pillars: storage, integration, governance, and access control. These four pillars of data architecture assume data will be centralized or federated through replication.
Zero-copy collapses the integration layer. If you never move data, you never need to reconcile copies. You never need to rebuild pipelines when source schemas change. You never need to answer the question, "Which copy is the current one?"
But governance and access control remain unsolved. A zero-copy query still requires knowing whether the record being queried is accurate, current, and appropriately consented. The architecture prevents duplication. It does not prevent bad data from being queried in place.
This is where health data trust without movement becomes the real design challenge. You need a trust layer that evaluates records at their source, before any query returns results.
What are the 5 pillars of zero trust?
Zero trust security architecture rests on five pillars: identity, device, network, application workload, and data. The model assumes that no entity inside or outside a perimeter should be trusted by default. Every access request must be verified.
Zero-copy and zero trust share a philosophical foundation. Both reject default trust. Zero trust says, "Do not assume the user is authorized." Zero-copy says, "Do not assume the data needs to move." Combined, they create a model where queries execute against verified data, by verified users, without any record leaving its source.
The zero trust model relates directly to application security because it enforces verification at every layer of the stack. In healthcare, this means verifying not just who is asking for data but whether the data itself meets a defined trust threshold before it enters a model or a clinical workflow.
Why zero-copy still needs trust scoring
A zero-copy query against an EHR that contains stale, unconsented, or poorly sourced records returns stale, unconsented, or poorly sourced results. The architecture is clean. The data is not.
SuperTruth's Data Trust Index scores every health data record from 0 to 100 across eight dimensions: Provenance at 25%, Consent at 20%, Recency at 15%, Quality at 10%, Concordance at 10%, Validation at 10%, Breadth at 5%, and Stability at 5%. This scoring happens at the point of origin, before any query or model touches the record.
In a zero-copy architecture, DTI scoring functions as a pre-query filter. A research consortium can set a DTI floor of 70 and ensure that only records meeting that threshold are queryable. No data moves. No copies are created. But every result that comes back carries a verified trust score.
This is the missing piece in current zero-copy implementations. Salesforce, Snowflake, and Databricks solve the plumbing. Nobody scores the water.
Key statistics
SuperTruth processed 105,000 diagnostic records for imaware, reducing standardization time from 3 weeks to 2 hours. That is a 95% reduction in processing time.
The imaware partnership saved over 200 hours per month in manual data operations and identified a previously invisible patient segment driving 20% of revenue.
The average health system maintains 5 to 7 copies of each patient record across its analytics, research, and operational environments. Each copy introduces an independent drift vector.
Healthcare data breaches cost an average of $10.93 million per incident in 2023, according to IBM. Every unnecessary copy of a health record increases breach surface area proportionally.
DTI's Provenance dimension, weighted at 25%, is the single most influential factor in determining whether a record is trustworthy enough for clinical AI or regulatory submission.
The architecture that actually works
The target state is a zero-copy, trust-scored health data architecture. Records stay at their source. Trust scores are computed at the source. Queries execute against scored data. Results include both the requested clinical information and the trust metadata that describes its reliability.
This model eliminates the two biggest failure modes in health data: unnecessary movement and unverified trust. It satisfies zero trust security requirements because access is verified at every layer. It satisfies regulatory requirements because provenance and consent are scored before data enters any downstream workflow.
Federated learning networks have attempted something similar, but they still carry trust problems because federation does not inherently verify the quality or provenance of the data being federated. Zero-copy with DTI scoring closes that gap.
For health systems preparing for AI deployment, the question is no longer how to move data safely. The question is how to score data where it lives so that movement becomes unnecessary.
SuperTruth's Clean Rooms and DTI floor enforcement let research consortia query across institutions without exposing individual records. If your team is managing federated data or clinical trial supply, contact Louis Simeonidis at louis@supertruth.ai or (215) 918-4140.
Further reading:
Jason Alan Snyder
Co-founder of SuperTruth and Artists & Robots, and an inventor on the Data Trust Index patents. Twenty-plus years building technology inside Interpublic Group. He writes here nearly every day on data trust, provenance, and what AI should be allowed to act on, and publishes essays on his Substack.
About SuperTruth · LinkedIn · Substack · jasonalansnyder.com
See it in practice
DTI scores the record, not the patient.
8 dimensions. 0–100. Travels with every record permanently.