Trust and governance

Grounded in real work,
built to be inspected.

We earn trust the same way we build evidence: by being precise, by showing our work, and by keeping raw data where it belongs. This page states plainly how data is handled, how the method is validated, and which guidance we answer to.

01What holds it up

Four things a careful reader can check.

HIPAA compliant

Built to the standard your data already lives under.

De-identification, access control, and provenance are handled inside your existing HIPAA framework. We operate within it, we do not ask you to work around it.

Regulatory aligned

Methods that answer to published guidance.

Cohort assembly and endpoint derivation follow FDA guidance on externally controlled trials and on AI in regulatory decision-making, with EMA precedent behind the approach.

Clinical partnerships

Grounded in real work with real institutions.

Active data partnerships with clinical partners in the US and India. Per our naming rule, we describe partners by type, not by name, until they choose otherwise.

Item-level traceability

Every endpoint traces back to its source record.

A derived scale is never a black box. Each item links to the exact sentence, value, or measurement it came from, so a biostatistician can inspect the provenance of every point.

02The de-identification method

Data is de-identified at source, and structured in place.

No raw record leaves the institution that holds it. Here is what that means step by step, in plain language.

  1. 01

    De-identified at source

    Identifiers are removed before any processing begins, inside the partner's environment. Sapien never receives an identifiable record.

  2. 02

    Structured in place

    The de-identified data is structured and harmonized where it already lives. It is analysis-ready on the partner's own servers, not on ours.

  3. 03

    Accessed through a governed key

    Sapien works through a governed research key with defined access and provenance. Governance is agreed with the partner before any data moves.

  4. 04

    Only evidence crosses the boundary

    What leaves is a de-identified, structured evidence package, traceable back to its source for audit. Raw records never leave the partner's environment.

DPDP and GDPR position

Because identifiers are removed at source and raw records stay inside the partner’s governance boundary, processing sits within existing HIPAA, GDPR, and India’s DPDP frameworks rather than working around them. Governance, access, and provenance are defined with the partner before any data moves. Every output can be traced to a source record for audit, while the identifiers that would make it personal data were never present in what Sapien handles.

03Regulatory anchors

The guidance the method answers to.

Public regulatory references, not our partners. These frame how an external control arm is designed, validated, and defended.

FDA · 2023 draft

Externally Controlled Trials for Drug and Biological Products

The FDA draft guidance on the design and conduct of externally controlled trials. The reference frame for how an external control arm is built and defended.

FDA · Jan 2025 draft

Use of AI to Support Regulatory Decision-Making

Introduces a seven-step credibility-assessment framework for AI used in regulatory submissions. Our validation and traceability are designed against it.

EMA · 2022

PROCOVA qualification

EMA qualified a covariate-adjustment method for sample-size reduction in Phase 2 and 3 trials with continuous outcomes. Precedent that regulators engage with model-based control methods.

ICH · E10

Choice of Control Group in Clinical Trials

The international guideline on selecting a trial control group, including external controls. The long-standing basis for when an external comparator is appropriate.

The best way to trust it
is to inspect it.

We would rather walk your team through the provenance of a single endpoint than make a claim you cannot check. Book a working session and bring your hardest question.