Skip to main content
Provenance & Explainability
Can you prove every conclusion?

Don't trust us.
Verify us.

Every regulated conclusion generated by PrismCDM carries a continuous provenance chain from source to inspection. From the original protocol through AI assessment, operational execution, evidence generation, regulatory submission, and inspection.

Provenance is not another report. It is the thread connecting every artifact PrismCDM produces.

Intelligence is explainable. Execution is traceable. Submission is defensible. One platform. One evidence model. One provenance chain.

Analytical Provenance

How PrismCDM reaches every conclusion.

The graph below shows the structural chain from public protocol source through engine extraction and assessment to the recommendation surface. Every transition is versioned, timestamped, and attributable to its originating source.

Provenance Chain
NCT02******

Every recommendation traces to its protocol source, engine version, evidence record, and build. The graph below shows the structural chain from public protocol source to PrismCDM's analytical output.

1
Source
Public Source
ClinicalTrials.gov

Public protocol record retrieved from ClinicalTrials.gov for the demonstration NCT identifier. Structured fields include sponsor, phase, eligibility criteria summary, primary outcome, and study design.

2
Extraction
Prism Generated
Clinical Intelligence Engine

The Clinical Intelligence Engine extracts structured intelligence from the public protocol record: complexity dimensions, operational burden, eligibility characteristics, schedule and visit patterns, endpoint structure. Engine version stamps every extraction.

3
Assessment
Prism Generated
PTII™

The Prism Trial Intelligence Index composite + its sub-indices (Protocol Complexity, Operational Burden, Eligibility Strictness, Amendment Risk, Enrollment Velocity) are computed from the extracted intelligence. The assessment methodology version stamps every score.

4
Finding
Prism Generated
FIND-*

Findings are structured observations the engine surfaces from the extraction + assessment layers. Each finding identifies a specific operational, eligibility, or scientific consideration and references the source layer that produced it.

5
Recommendation
Prism Generated
REC-*

Recommendations connect findings to concrete operational actions. Each recommendation cites the findings that supported it and carries a priority + expected impact. Every transition in this chain is logged with engine version, assessment version, corpus snapshot, and build hash, recorded at the moment of generation.

Verifiable. Reproducible. Inspection-ready.

Demonstration generated by PrismCDM using publicly available ClinicalTrials.gov protocol information (NCT02******). This analysis was independently produced by PrismCDM and has not been reviewed or endorsed by the study sponsor.

Operational Provenance

How that conclusion propagates through regulated execution.

Every recommendation issued by the Clinical Intelligence Engine propagates through operational decisions, evidence records, and regulatory artifacts. The five-node chain below shows how PrismCDM preserves the audit trail from the moment a recommendation is issued to the moment an inspector reviews the submission package.

Operational Provenance Chain

From recommendation to inspection. Every step is logged with the operator that performed it, the timestamp, and the regulated artifact it produced.

1
Recommendation

The Clinical Intelligence Engine issues a structured recommendation citing the findings that supported it. Each recommendation carries priority + expected impact.

2
Operational Decision

The recommendation translates into an operational decision: site selection, screening cadence, monitoring focus, schedule adjustment. The decision is recorded with the operator that made it and the audit-trail entry that justifies it.

3
Evidence Record

Every operational event accumulates into the Inspection Evidence Package. The decision above becomes a section entry tied back to the recommendation that triggered it.

4
Submission Artifact

The accumulated evidence rolls up into the regulatory submission package. The recommendation is now a permanent part of the audit-defensible record.

5
Inspection

An inspector reviewing the submission can trace any conclusion in the package back through the chain above: from the inspection record to the submission artifact to the evidence section to the operational decision to the recommendation, and from there into the analytical provenance chain that produced the recommendation in the first place.

Versioned. Timestamped. Attributable.
Explainability vs Provenance

Two distinct questions. One platform answers both.

Explainability

“Why did the AI reach this conclusion?”

Scope
Intelligence
Timing
Before execution
Purpose
Supports scientific review

Provenance

“Where did this conclusion originate, and how did it propagate?”

Scope
Entire platform
Timing
Throughout execution
Purpose
Supports regulatory inspection
Why The Status Quo Is Insufficient

Why clinical development needs a new operating system.

Clinical execution is fragmented.

Protocols, execution, evidence, and regulatory preparation live in disconnected systems. The work to keep them in sync absorbs the team's time.

AI without provenance is difficult to trust.

Clinical teams need to understand not only what AI recommends, but how it reached every conclusion. Without that, regulators and reviewers cannot accept it.

Evidence should be generated continuously, not assembled at the end.

Inspection readiness should emerge from daily operations, not become a last minute project before the audit window closes.

The Platform Substrate

One Platform. Four Capabilities.

Every artifact is produced by the same platform that powers Clinical Intelligence, Execution, Evidence, and Provenance across PrismCDM. The Brief is not a feature. It is proof of the platform.

Clinical Intelligence

Understand every protocol before execution begins.

Structured findings on complexity, operational burden, eligibility, schedule, and endpoints. Every finding is identified, scored, and traceable.

Execution

Transform protocol intent into operational decisions.

Recommendations connect each finding to a concrete next step. Site selection, screening reach, visit cadence, monitoring focus.

Evidence

Produce inspection-ready evidence continuously.

Every Brief, every workflow event, every reconciliation step generates the regulated evidence inspection teams expect, not assembled at the end.

Provenance & Explainability

Trace every regulated conclusion back to its source.

Engine version, assessment version, build hash, source layer, evidence anchor. Every finding and recommendation is reproducible and verifiable.

From Protocol to Submission

One platform across every deliverable your team scopes today.

The deliverables clinical teams actually build, in the order they build them. PrismCDM produces each one continuously as the study runs, with source-backed provenance and regulated audit history preserved end to end.

  1. Protocol
  2. ICF
  3. EDC
  4. SDTM
  5. ADaM
  6. SAP
  7. TFLs
  8. CSR
Produced by four platform capabilities
Clinical Intelligence
Execution
Evidence
Provenance & Explainability

Every deliverable, the same platform, the same provenance chain.

Who PrismCDM Is Built For

One platform across the entire clinical development sponsor stack.

Sponsors

Standardize how protocol risk is understood across every study in the portfolio.

Biotech

Run the same enterprise-grade Brief on every protocol with a lean clinical team.

CROs

Bring inspection-ready evidence and provenance to every sponsor engagement.

Academic Medical Centers

Generate Briefs for investigator-initiated protocols with the same rigor as industry.

Mid-size Pharma

Operate clinical development with the platform discipline of top-twenty sponsors.

Built for Regulated Clinical Development

Platform Principles

  • Multi-tenant
  • Append-only audit
  • 21 CFR Part 11
  • Role-based access
  • End-to-end provenance
  • Independently versioned engines
Early Access

See the chain back to its source.

Bring one of your protocols. We'll trace one finding from protocol through intelligence, execution, evidence, submission, and inspection so you can see every decision, every transformation, and every audit record along the way.