Service · Statistical Analysis Plan
Statistical Analysis Plans for wearable & digital endpoints
A wearable SAP lives or dies on decisions made before a single model runs: how the endpoint is defined, what counts as a valid wear day, and how missing data is handled. We write SAPs that pre-specify those choices defensibly, so the analysis survives review instead of being rebuilt after it.
What a wearable SAP must nail
The decisions that determine the result
Endpoint definition
Precise, defensible definitions for the chosen sensor metrics — what is measured, over what window, and how it is aggregated.
Wear-time & valid-day rules
Pre-specified compliance thresholds and valid-day criteria, with the evidence base behind them — the choice that quietly decides whether the endpoint is powered.
Estimands (ICH E9(R1))
Treatment effect framed with the estimand structure reviewers now expect, including how intercurrent events and non-wear are handled.
Primary model
MMRM / mixed-effects specification, covariates, and handling of repeated daily/longitudinal measures.
Missing-data strategy
A characterized missingness mechanism and a matched handling approach (not complete-case by default), with a sensitivity battery.
Reliability & interpretation
Where relevant, ICC/MDC and MCID framing so a detected change is both real and meaningful.
Two ways to engage
We can write it, or review yours
Already have a SAP drafted? We also provide independent SAP review and commentary: a sensor-data specialist marks up the wearable-endpoint sections (endpoint definition, wear-time and inclusion/exclusion rules, the primary model, missing-data handling) and flags what a reviewer would question before it locks.
Why pre-specification matters
Get it right before the data locks
It's hard to write a good SAP around failure modes you haven't met: unexpected data issues, compliance drift, and the pushed deadlines that follow catch most teams on their first wearable endpoint. We've seen where the data goes wrong, and we plan for it up front. Pre-specifying the wear-time threshold, the estimand, and the sensitivity battery is what turns a promising sensor signal into a regulatory-grade endpoint. This work pairs naturally with platform-specific analysis for APDM Opal, Garmin & Fitrockr, and actigraphy.
Planning a trial with a wearable endpoint?
Best engaged before the SAP locks. A short call scopes the endpoint, the data, and the plan.
Get in touch