Device data on Centralive

Centralive passively collects research-grade data from wearables, phones and connected devices across four providers. Available metrics depend on the participant’s device model and the study’s configuration — and every export is de-identified.

    Garmin · Oura · Withings · AppleSecure OAuth 2.0De-identified CSV
    Supported devices

    Pick a provider.

    Pick a provider to browse its device types as tabs, or open Compare devices to line two or three up side by side.

    At a glance

    ProviderDevice familiesHeadline data
    GarminWatches · Index BPM · Index S1/S2 scalesHeart rate, HRV, activity, sleep, stress, SpO₂, body composition, blood pressure
    OuraRing Gen 2 / 3 / 4 / 5Sleep, readiness, HRV, heart rate, SpO₂, activity, temperature, cycle insights
    WithingsBPM · scales · ScanWatch · Sleep Analyzer · Thermo · BeamOBlood pressure, ECG/AFib, body composition, sleep, SpO₂, temperature, vitals
    AppleApple Watch · iPhone · AirPods (SensorKit, custom study app)Heart rate, ECG, PPG, motion, sleep, wrist temperature, phone and app use, speech and typing metrics, places, ambient light
    Help Center / Tools / Compare

    Compare devices

    Line up two or three devices to see which metrics each one captures — Garmin watch tiers against each other, or a Garmin watch against the Withings ScanWatch.

    1Choose a device type
    2Add devices to compare

    High level shows the broad data types each device captures (the number is how many individual metrics it provides in that group); switch to Detailed for every metric. Garmin watches are grouped by feature tier, so capabilities are indicative — a few metrics (e.g. ECG, SpO₂ or skin temperature) can vary by exact model, firmware and region. Rows where the devices differ are highlighted.

    Exports

    How exports work

    Exported data is grouped by device class, then by category, then by participant. Every data file shares the same leading columns — participant ID, local timestamp, timezone and UTC timestamp — followed by the metric values and the device model. Free-text and account identifiers are dropped at ingestion, so exports stay de-identified.

    participant_idtimestamp_localtimezonetimestamp_utcmetric values …device_model