The DEEP Stack model was designed to structure digital measures from concept through to clinical evidence. But how well does that structure translate into the data standards required for regulatory submission? We put it to the test.
The Challenge
Digital measures collected by wearable devices generate data types that existing clinical data standards were never built for. Continuous sensor streams, algorithm-derived endpoints, and device hierarchies don't map neatly to traditional tabulation models. The result: every sponsor reinvents the wheel, mapping their digital endpoints to CDISC standards from scratch - a costly, error-prone process that slows regulatory submissions and limits cross-study comparability.
What We Did
We selected Stride Velocity 95th Centile (SV95C) - the first wearable-derived digital measure with EMA regulatory acceptance as a primary endpoint - and mapped it against the CDISC standards chain: USDM for protocol and endpoint definitions, SDTM for data tabulation (device domains and findings), and ADaM for analysis-ready datasets. The CDISC DHT Portal's Step Count example served as the closest published reference pattern.
This work was accepted for poster presentation at CDISC Interchange Milan 2026.
What We Found
Of the 20 data elements assessed, 8 mapped cleanly to existing CDISC structures - device identifiers, parent-child hierarchies, result values, and temporal variables all transferred directly. Four mapped partially, requiring controlled terminology extensions. Five were gaps: no published CDISC test codes, method terms, or device type classifications exist for this class of wearable-derived gait measure. Three areas represent genuine extensions where the DEEP Stack captures information - such as technical solution requirements and algorithm transparency - that CDISC standards do not yet address.
From this single measure, we identified three Controlled New Terminology (Cnew) candidates, each backed by an EMA-accepted use case - ready to feed back into CDISC governance.
Why This Matters
For pharma sponsors: Every digital endpoint your trials deploy faces the same standards gap. This proof of concept demonstrates a systematic approach to identifying what maps, what's missing, and what needs to be proposed to CDISC - before it becomes a submission bottleneck.
For technology companies: If you build wearables or algorithms that generate clinical data, your customers need that data in SDTM and ADaM. Understanding how device metadata, sensor specifications, and algorithm outputs align with CDISC domains is essential for positioning your technology as trial-ready.
For CROs: You manage the data pipeline from collection to submission. A structured, repeatable approach to mapping novel digital measures to CDISC standards reduces programming time, minimises rework, and strengthens the traceability regulators expect.
Partner With Us
A prototype of the DEEP Trusted Regulatory System exists. We are now seeking industry partners to co-develop and pilot the full standards chain - from measurement concept definition through CDISC-mapped tabulation to analysis-ready datasets.
The pilot is scoped as a six-month, three-phase programme covering USDM protocol mapping, SDTM domain mapping, and ADaM dataset generation, with CDISC engagement at each stage. We're looking for pharma sponsors, technology companies, and CROs who want to solve the digital data standards problem together rather than in isolation.
For inquiries regarding participation or additional information, please contact us here.