The truth score ceiling: why Platinum-grade data cannot come from unverified sources
Unverified health data hits a hard ceiling in the Data Trust Index. No amount of recency, breadth, or quality improvement can push a record past Gold grade if its provenance chain includes unverified sources. The DTI Platinum ceiling is not a bug. It is the entire point of trust-scored health data.
Platinum-grade health data cannot be manufactured from unverified sources. This is not a theoretical claim. It is a mathematical constraint built into the Data Trust Index scoring architecture. A record sourced from an unverified origin faces a hard ceiling that no amount of optimization across other dimensions can overcome.
The current SERP for platinum-related queries focuses on precious metals, mining deposits, and commodity pricing. None of it addresses the concept that matters most to healthcare AI: the structural ceiling that prevents low-provenance data from reaching the highest trust tier. That gap is the subject of this post.
What the DTI Platinum ceiling actually means
The Data Trust Index scores every health data record from 0 to 100 across eight dimensions. Provenance carries the heaviest weight at 25%. Consent follows at 20%. Together, these two dimensions account for 45% of a record's total score.
Platinum grade requires a DTI score of 90 or above. A record that scores zero on provenance, meaning its source is entirely unverified, loses 25 points immediately. Even if that record scores perfectly across every remaining dimension, its maximum possible score is 75. That puts it squarely in Silver territory, two full tiers below Platinum.
This is the DTI Platinum ceiling. It is not a soft penalty. It is a structural limit.
Why provenance is weighted at 25%
Provenance answers one question: where did this data come from, and can we verify that chain of custody?
A lab result from a CLIA-certified laboratory with full LOINC coding, a traceable order from a credentialed provider, and a timestamp from a validated EHR system scores high on provenance. The same lab value copied into a spreadsheet by an unknown party, forwarded through email, and uploaded to a data lake without metadata scores near zero.
The information might be identical. The trust is not.
SuperTruth weights provenance at 25% because every downstream use of a health data record depends on knowing its origin. An FDA submission requires chain of custody. A clinical decision support tool needs to know whether the glucose reading came from a validated device or a patient's handwritten note. A payer adjudicating a prior authorization needs to know whether the diagnosis code was assigned by the treating physician or inferred by an algorithm.
Without verified provenance, none of these use cases can operate at the highest confidence level. The 25% weight reflects that reality.
The four DTI trust tiers
The Data Trust Index assigns every scored record to one of four tiers:
The ceiling imposed by unverified provenance means a record cannot cross from Silver to Gold, let alone reach Platinum, without resolving its source verification gap.
Why unverified health data trust has hard limits
Some organizations treat data quality as a continuous improvement problem. Add better validation, improve coding accuracy, increase breadth of capture, and the data gets better over time. That logic works within a tier. It does not work across the provenance boundary.
Consider a dataset of 50,000 patient records ingested from a digital health app. The app collects symptom logs, medication adherence timestamps, and biometric readings from consumer wearables. The data is recent, broad, well-structured, and internally consistent. It scores well on recency (15%), quality (10%), breadth (5%), and stability (5%). It might even have strong consent documentation through the app's terms of service.
But the biometric readings come from consumer-grade devices without FDA clearance. The symptom logs are self-reported without clinical validation. The medication timestamps reflect app interactions, not verified dispensing events. Provenance scores low because no verified clinical source exists in the chain.
That dataset hits a ceiling. It can reach Gold grade with strong performance across non-provenance dimensions. It cannot reach Platinum. The data trust grade limits are structural, not aspirational.
How this differs from precious metal grading (and why people ask)
People searching for platinum-related terms often ask: why is platinum not as popular as gold? The answer in commodity markets involves supply dynamics, industrial demand ratios, and investor sentiment. In health data trust, the analogy is different but instructive.
Gold-grade health data is more common than Platinum-grade for the same reason gold bullion circulates more freely than platinum: the barrier to entry for Platinum is structurally higher. Most health data originates from systems with incomplete provenance chains. EHR data crosses multiple systems during care transitions. Claims data passes through clearinghouses that strip metadata. Patient-reported data enters through consumer interfaces without clinical validation.
Gold grade is achievable with strong performance across multiple dimensions and adequate provenance. Platinum requires verified provenance at every step in the chain. That requirement eliminates most health data from Platinum eligibility before any other dimension is evaluated.
Another common question: is diamond higher than platinum? In gemstone marketing, yes. In data trust scoring, no tier exists above Platinum. A score of 100 is theoretically possible but rare in practice. It would require perfect scores across all eight dimensions, including verified provenance from a single, validated clinical source with full chain of custody documentation.
How to identify raw platinum (in data terms)
In metallurgy, identifying raw platinum requires specific gravity tests, streak analysis, and acid resistance evaluation. In health data, identifying Platinum-grade candidates requires a different but equally rigorous assessment.
Platinum-eligible data typically has these characteristics:
Most health datasets fail at least one of these criteria. The DTI Engine identifies which criteria fail and by how much, giving organizations a precise remediation path.
Is there a shortage of Platinum-grade data?
Yes. The shortage is severe.
When SuperTruth scored 105,000 diagnostic records for imaware, the process revealed systematic gaps in provenance documentation that would have prevented large portions of the dataset from reaching Platinum without remediation. The scoring process itself, which reduced standardization time from 3 weeks to 2 hours, was the mechanism that identified where provenance chains broke.
This shortage is not unique to imaware. It reflects a structural reality across healthcare. Most health data systems were built for billing, not for trust verification. The infrastructure that generates health data does not, by default, preserve the metadata required for Platinum-grade provenance.
The shortage creates a competitive advantage for organizations that invest in trust-scored data infrastructure early. As the FDA increases scrutiny of AI training data provenance, and as CMS programs like ACCESS require auditable data foundations, Platinum-grade data will become the differentiator between compliant and non-compliant AI deployments.
What is rarer than Platinum-grade data?
Platinum-grade data that maintains its score over time. Data trust is not static. A record that scores 92 today may score 84 in six months if its recency dimension degrades, if the consent that authorized its use is revoked, or if the provenance chain is disrupted by a system migration.
This is why SuperTruth treats the DTI score as a living measurement, not a one-time certification. The DTI score as a contract defines what Platinum-grade data guarantees at the moment of query. Maintaining that guarantee requires continuous scoring.
Key statistics
The consent multiplier on the provenance ceiling
Provenance is not the only dimension that imposes a ceiling. Consent, weighted at 20%, creates a second structural barrier for unverified data.
Data from unverified sources frequently has consent gaps. If you cannot verify where data came from, you often cannot verify whether the consent obtained at the point of collection covers the intended downstream use. A digital health app may have collected broad consent for "service improvement" but not for clinical AI model training or FDA regulatory submission.
When both provenance and consent score poorly, the ceiling drops further. A record scoring zero on both dimensions loses 45 points, capping its maximum score at 55. That is Bronze grade. No amount of data quality, recency, concordance, or validation can rescue it.
This dual ceiling is why organizations working with unverified health data trust face compounding limitations. The problem is not that one dimension is weak. The problem is that unverified sourcing corrupts the two highest-weighted dimensions simultaneously.
Why the ceiling matters for AI model training
AI models trained on Silver-grade data produce Silver-grade outputs. This is not a metaphor. If a sepsis prediction model trains on data where 30% of records have unverified provenance, the model inherits that uncertainty. It cannot distinguish between a verified vital sign from an ICU monitor and an unverified reading from an unknown device.
The FDA's emerging guidance on AI/ML training data provenance points toward exactly this problem. Regulatory reviewers will ask: what was the trust grade of your training data? If the answer is "we did not score it," the follow-up question is: then how do you know your model's outputs are trustworthy?
The DTI Platinum ceiling provides a clear, auditable answer. Organizations that enforce a minimum trust tier for AI training data can demonstrate to regulators that their models were built on verified foundations. Organizations that skip this step will face increasing regulatory friction as FDA enforcement matures.
Remediation paths: moving data toward Platinum
The Platinum ceiling is hard but not permanent. Data that currently scores in Silver or Gold can be remediated toward Platinum through specific interventions:
Not all data can be remediated. Some provenance chains are irretrievably broken. The DTI scoring system identifies these cases explicitly, allowing organizations to make informed decisions about which data to invest in remediating and which to exclude from high-stakes use cases.
The competitive implication
Organizations that understand data trust grade limits will build different data strategies than those that do not. If you know that unverified data cannot reach Platinum, you stop investing in downstream optimization of unverifiable datasets. You redirect resources toward source verification and provenance infrastructure.
This shift has measurable ROI. imaware's experience demonstrated that trust-scored data processing identified the customer segment driving 20% of revenue, a finding that was invisible in unscored data. The DTI scoring process did not just clean data. It revealed which data was worth trusting and which was not.
The Platinum ceiling is not a limitation of the scoring system. It is the scoring system working as intended. Platinum-grade data cannot come from unverified sources because the entire value of Platinum grade is the guarantee that it did not.
The DTI Engine scores every health data record 0-100 across 8 trust dimensions before your AI model sees it. If your team is evaluating data for training, compliance, or clinical use and needs to understand where your data hits its trust ceiling, 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.