Clinical & regulatory rigor.
Coordinated execution.
Execute clinical validation against your defined intended use, reference standard and protocol—with data, models, clinical experts and traceable results in one workflow.
Your study protocol.
Executed end to end.
You bring the study design, intended clinical use and reference-standard requirements. We coordinate and execute the agreed work with the right data, model versions and clinical experts.
- 01
The right data.
Source, license, curate and manage cohorts against the study’s inclusion criteria and data requirements.
- 02
The right model version.
Register the customer model, configure its inputs and outputs, and generate inference on the selected cohort. Preserve the version and execution settings.
- 03
Clinical annotation & adjudication.
Coordinate US board-certified physician annotators, review and adjudication against the agreed reference-standard protocol. Where appropriate, AI pre-annotations are reviewed and corrected by experts.
- 04
Protocol execution & analysis.
Compare model outputs with adjudicated expert judgment and quantify performance according to the agreed scientific and regulatory evaluation protocol.
- 05
Traceable evidence.
Connect results to the model version, dataset, annotations and analysis settings, with the records needed for review and subsequent evaluation.
Clinical annotation: expert image review and reference labels, connected to the study.
Run the next version.
Preserve the comparison.
Revalidate a model update against the agreed cohorts, protocol and expert reference. Run inference, compare performance with the prior version, and retain the data, model, annotation and analysis context behind the result.
Revalidation can support updates with or without a PCCP. When an FDA-authorized PCCP applies, we execute the evaluation within its specified protocol. Your regulatory team determines the change pathway, authorized scope and submission requirements.
See models and versions in action →Monitor performance.
Keep the evidence connected.
Run post-market monitoring in live-use settings, with the agreed data feeds, measures and review responsibilities. Keep model version, cohort context and performance evidence connected as signals lead to investigation, expert review or revalidation.
Our monitoring module brings statistical methods from FDA’s public AI-monitoring research into the same environment as data, models, annotation and validation. Trace findings and follow-through within your quality and change-control processes.
Clinical-performance measures depend on suitable outcome or reference data. Input drift alone does not establish a change in clinical performance.
Explore our monitoring approach →Plan the change.
Execute the evidence.
An AI-device PCCP describes planned modifications, the methodology for developing, validating and implementing them, and an assessment of their impact.
A PCCP must be authorized by FDA before changes are implemented under it. A change within that plan must follow its specified modifications, validation methods and acceptance criteria. Changes outside the plan require their own regulatory assessment.
Gesund executes the agreed clinical evaluation and preserves the evidence. Study design, regulatory advice and submission decisions remain with your team and its advisers.
Read FDA’s final PCCP guidance ↗A PCCP is not a prerequisite for revalidation and does not remove validation or post-market responsibilities.
From your protocol.
To traceable evidence.
Align the execution
Confirm the supplied protocol, inputs, responsibilities and deliverables before execution begins.
Run the study
Coordinate data, model inference, physician annotation, adjudication and performance analysis through one managed workflow.
Preserve the evidence
Deliver auditable, version-linked results and their supporting records, with agreed access, exports and retention.
Gesund executes clinical validation and evidence workflows. Study design, regulatory strategy and the manufacturer’s quality system remain with your team and advisers. Our service does not replace a complete QMS, cybersecurity testing or all software/device validation, and does not guarantee regulatory authorization.
What does your next study need to prove?
Bring your product, protocol and clinical evidence question. Let’s put the work in motion.