Harmonization layer

Vendor-agnostic

Apple, Fitbit, Garmin, Withings, Oura, Polar, ActiGraph

Latency

Real-time

Normalized as the signal arrives

Output

Endpoint-ready

Aligned to protocol windows and study logic

Wearable harmonization built for real-world trials

Real-Time Signal Harmonization
for Clinical Trials

Signal harmonization is not simply pulling data from a wearable API. It is normalizing heart rate, HRV, SpO₂, sleep, steps, respiratory and activity signals from multiple device sources into one consistent clinical data model — in real time, with QC context and study-ready outputs that protect digital endpoint integrity.

Vendor-agnostic · Endpoint-aligned · Real-time

UNITS
TIME
QC
MODEL

Harmonization Workflow

Ingest → Normalize → QC → Endpoint

Why is my Apple Watch heart rate so different from the Fitbit?
Each device exports different units, sampling rates, and timestamps. Harmonization is what makes them comparable for clinical analysis.
The goal is not just ingestion. It is a clinical data model that holds together over a multi-year study.
Normalize the signal, protect the endpoint Units + time + QC + clinical model

What Signal Harmonization Really Means in Clinical Research

In clinical trials, signal harmonization is the layer that transforms raw wearable telemetry into a clinically consistent dataset. It is what allows a study to treat heart-rate data from a Garmin and an Apple Watch as the same measurement type — without losing the device-specific context that affects how that measurement should be interpreted.

Harmonization is what turns multi-vendor wearable data into a single analyzable stream — without erasing the provenance regulators expect to see.

Related pages: Wearables · Signal QC

Multi-device wearable signals being harmonized into one clinical data model

Why Signal Harmonization Breaks in Trials

Most trials don't notice harmonization is broken until analysis. By then, the dataset already has problems that can't be undone — only documented.

Mixed units

Beats per minute, milliseconds, percent, and steps from different vendors look the same column but mean different things.

Drifting timestamps

Devices and apps use different clocks. Without time normalization, comparing across devices produces ghost trends.

Inconsistent sampling

One device emits a sample per minute, another per second. Without harmonization, summaries become apples-to-oranges.

Hidden QC state

Vendor flags for motion artifact, low confidence, or sensor disconnect get dropped during ingestion.

Vendor model drift

Algorithm updates from a vendor change how a metric is computed — silently, mid-study, without versioning in the data.

Endpoint mismatch

Harmonized output that doesn't match how the protocol defines a valid day or valid window forces post-hoc analysis.

Harmonization isn't a one-time ETL job. It's an operational layer that has to be versioned and monitored over the life of the study.

Delve Harmonization vs Raw Vendor Streams

Raw vendor streams

  • Each device speaks its own format
  • No unified time base across vendors
  • QC flags lost during ingestion
  • Silent algorithm changes mid-study
  • Hard to define a 'valid day' across devices

Delve harmonized model

  • One clinical data model across vendors
  • Aligned timestamps and sampling
  • QC flags preserved and visible
  • Versioned algorithms with change history
  • Protocol-aligned valid-day / valid-window logic

Harmonization is what makes a multi-vendor wearable trial behave like a single coherent dataset.

What a Strong Harmonization Model Includes

Harmonization that survives a multi-year, multi-device trial is built around the lifecycle of the signal — not just the day-one connection.

A strong harmonization layer means the same patient signal can be interpreted the same way from day one through the last visit — across whatever devices the protocol allows.

See related pages: Devices · Analytics · Support

Multi-vendor wearable signals harmonized into one clinical data model with QC context

Harmonization Enables Digital Endpoint Strategy

Digital endpoints can only be trusted when the underlying signals are comparable across patients, devices, and time. Delve builds harmonization around endpoint validity — not just data transport.

Protocol-aligned models

Harmonized outputs map to the study's own definitions of valid measurements, windows, and endpoints.

Cross-device comparability

Two patients on two different wearables can still be analyzed against the same digital endpoint.

Algorithm version control

Every endpoint is tied to a specific algorithm version, so results stay reproducible.

QC-aware analytics

Downstream dashboards and analyses know which datapoints carried vendor-flagged QC issues.

Faster recovery

When a harmonized signal drifts or stops, the operations team sees it within hours — not at next interim analysis.

Submission readiness

Harmonization documentation supports the lineage and provenance regulators expect for digital endpoints.

Harmonization is one of the core infrastructure layers behind scalable, defensible digital endpoint execution.

FAQ

Is an API integration enough to harmonize wearable data?

No. An API gives you the raw stream. Harmonization is the additional layer that aligns units, time, sampling, QC, and endpoint definitions so the data is comparable across vendors and patients.

Can Delve harmonize signals from devices we already use?

Yes. The model is vendor-agnostic and supports the major wearable platforms used in clinical research. New device types are added through the same harmonization pipeline.

How does harmonization affect data submission?

Harmonized outputs preserve provenance — device, firmware, algorithm version, QC flags — so the dataset is defensible in a regulatory review and can be re-analyzed if needed.

Build Multi-Device Wearable Trials That Hold Together

Delve combines real-time harmonization, signal QC, analytics, and participant support into one operational model designed to protect digital endpoint integrity across devices and time.

Book a Harmonization Discussion

Explore Delve wearables capabilities →