FAQ
Wearable biostatistics, answered
Common questions about turning wearable, IMU, accelerometer, and PPG data into defensible clinical-trial endpoints. Don't see yours? Get in touch.
What kinds of wearable data do you analyze?
We work across motion and cardiac signals: inertial measurement unit (IMU) and accelerometer data for gait, postural sway, and activity, and optical photoplethysmography (PPG) data for heart rate, heart rate variability (HRV), and RMSSD. That spans clinical-grade platforms like APDM Opal and Ametris (formerly Actigraph) through consumer wearables such as Garmin, Fitbit, Apple Watch, Oura, and WHOOP.
What makes a wearable endpoint "defensible"?
A defensible endpoint is one whose definition, wear-time rules, estimand, and analysis model are pre-specified before the data locks, with a documented evidence base behind each choice. We align deliverables with FDA digital health technology (DHT) guidance, the V3 framework (verification, analytical and clinical validation), and the ICH E9(R1) estimands addendum, the language reviewers and sponsors expect.
Can you derive HRV from raw sensor data?
Yes. We process raw beat-to-beat interval (BBI) data, the optical PPG-derived heart-rate signal, into HRV and RMSSD, including the artifact detection and correction that wrist-based PPG requires.
How do you handle wear-time and missing data?
Wear-time and valid-day rules quietly decide whether an endpoint is powered, so we pre-specify non-wear detection, the minimum valid-day threshold, and the evidence behind it. Missing data is handled with a characterized missingness mechanism and a matched approach plus a sensitivity battery, not complete-case by default.
Which devices and platforms do you support?
Deep experience with APDM Opal (IMU motion), Garmin and Fitrockr (PPG/BBI and accelerometry), and research accelerometers (Ametris (formerly Actigraph), activPAL, Axivity). We also analyze Empatica, Koneksa, Biosensics, WHOOP, Oura, Fitbit, and Apple Watch data. The methodology transfers across platforms.
Do you write Statistical Analysis Plans (SAPs)?
Yes. Reviewer-ready SAPs for wearable and digital endpoints, covering endpoint definition, wear-time and valid-day rules, the estimand (ICH E9(R1)), the primary model (typically MMRM or mixed-effects), and the missing-data strategy. Pre-specifying these is what turns a promising sensor signal into a regulatory-grade endpoint.
Do you only work with pharma, or also academic studies?
Both. We support pharma and biotech sponsors (Phase I–III), CROs needing specialist overflow capacity for sensor-data endpoints, and academic medical centers running investigator-led studies, with the same trial-grade, reproducible methodology in each case.
What software and standards do you work to?
Analysis is done in SAS and R, reproducible and validated, with deliverables aligned to FDA digital-health guidance. You receive analysis-ready datasets, statistical tables and figures, and manuscript-ready methods.
Can my team run the analysis with your support, instead of outsourcing it?
Yes, that is a core part of how we work. Alongside full-service analysis, we offer independent SAP review, independent validation programming, and the DIY Starter Pack: a fixed package that backs your in-house team at four checkpoints (SAP review, a training session, validation programming, and manuscript review). You keep the work in-house and build the capability while a sensor-data specialist makes sure it holds up.
How do engagements work, and what do they cost?
A short scoping call defines the problem, the data, and the deliverables. We engage on fixed-price projects, monthly retainers, time-and-materials, or prepaid hour blocks. Rates and structures are shared on request.
Is my data secure? Do you sign NDAs?
We work with de-identified data only and sign NDAs as standard. SOPs and references are available on request.
Have wearable data that needs to hold up?
A short scoping call defines the work and the price. No CRO overhead.
Get in touch