Digital biomarkers for program design

Quick answer: a digital biomarker is only as strong as the capture profile, wear-time plan, and export dictionary behind it. Design those with the endpoint, before kits ship.

A digital biomarker is a measured endpoint, not a dashboard tile

In research and program design, a digital biomarker is an endpoint you define and defend. The wearable is the capture instrument. If the protocol only stores someone else's recovery index, you do not have a biomarker you can re-open later. You have a screenshot with a timestamp.

Write the analysis question first. Name the feature (for example overnight HRV as rMSSD, respiratory rate from beat intervals, or activity seconds). Then ask whether the device will store the upstream signal or intervals you need to re-derive it.

Design the capture profile with the endpoint

Sample rate, duty cycle, and raw vs computed channels are biomarker design choices. They are not afterthoughts for IT.

For Sensor Bio programs:

• Sample rate is configurable per organization at 25, 50, or 100 Hz (default 50 Hz).
• Daytime Passive and overnight Rest Mode use different PPG duty cycles. Raw PPG is disabled by default in both. Turn it on before enrollment if morphology or re-derivation is in scope.
• On-device processing can produce heart rate, HRV (rMSSD), respiratory rate, steps, cadence, and beat-to-beat intervals. Mobile and cloud layers add sleep staging and trend scores. Know which layer owns the endpoint you claim.

If the endpoint only needs beat intervals and computed biometrics, say so and skip the power cost of raw waveform capture. If you may need optical morphology later, enable raw channels (green, and infrared/red when configured) while the cohort is still forming. See raw PPG access for the partner path, and how to configure a research wearable for profile setup.

Wear-time is part of the biomarker

An endpoint computed on empty nights is missing data, not a soft failure.

Build wear-time into the protocol the same way you build visit windows:

• Target hours worn per day and nights with valid sleep windows
• Charge and sync instructions participants will actually follow
• Who watches dropout after week one

Battery under default wellness settings is not battery under study settings. Sensor Bio V2 comparison figures list up to 7 days under default settings and up to 2 days under max recording settings, with much shorter on-device storage at max recording. High-rate raw capture needs a tighter charging SOP. Put that in the participant instructions, not in a footnote.

Put the data dictionary in the protocol

Before first pull, write down:

• Metric names and units (for example HR in bpm, HRV rMSSD in ms, skin temperature in °C)
• Whether each field is on-device, app, or cloud computed
• Export formats and windows (Sensor Bio: CSV/JSON; computed metrics 1-30 days; raw metrics 1-7 days)
• Automation path (REST at https://api.getsensr.io/v1/, iOS/Android SDKs)
• Who may download (Sensor Bio raw/computed downloads require Super Admin)

If your statistician cannot map columns to the SAP, the biomarker is not ready. Use the shorter research wearable checklist in vendor screens, and the wearables for clinical trials buyer guide when the RFP needs data-access language.

Boundaries that keep programs honest

Sensor Bio V2 is a general wellness device, not FDA cleared for diagnosis. Program copy should stay inside research and infrastructure framing. Do not treat disease labels, screening abstracts, or clinical-sounding marketing adjectives as product claims for the current device.

Keep thresholds and reference ranges out of public marketing unless legal clears them. Endpoint definitions belong in the protocol and analysis plan, not in a homepage tile.

Next steps

For day-one ops, freeze a wear-time planner, a device reset and reuse checklist, and a data dictionary before the first kit ships. Talk to engineering when you are ready to lock a device profile and export path. Book a platform walkthrough if the cohort timeline is already on the calendar.

Claims note

This article is for research and program design. It does not claim Sensor Bio V2 is FDA cleared, does not assert diagnostic performance, and does not imply RPM billing eligibility.