Mapping — Strava

Source: a Strava activity + its streams. Pillars: A + B. Mapper: mapStrava — mapStrava({ activity, streams }) → records[]. This is the cross-pillar shape end to end: telemetry becomes Pillar A Measurements, and a Pillar B Session references them via measuredBy rather than embedding them (§1.3).

Structural correspondence

Strava	OpenBody
streams.heartrate / watts / cadence	single-channel sampleArray Measurement (heart_rate /min, power W, cadence /min)
streams.latlng + altitude	one multi-channel location sampleArray (channels lat/lon/alt)
streams.time	the shared offsets array (irregular sampling)
average_heartrate / max_heartrate	interval quantity aggregates (heart_rate_mean/_max) with a derivedFrom link to the HR stream
the activity itself	Session (+ a continuous WorkUnit) with measuredBy links to every telemetry Measurement
device_name	provenance.device ({ manufacturer: "garmin", model }); sourceApp: "strava"

Input (excerpt)

{
  "activity": {
    "id": 14582931007, "type": "Run", "sport_type": "Run",
    "start_date": "2026-06-20T06:30:00Z", "elapsed_time": 2600,
    "moving_time": 2520, "distance": 10000.0,
    "average_heartrate": 152.0, "max_heartrate": 178,
    "device_name": "Garmin Forerunner 965"
  },
  "streams": {
    "time":      { "data": [0, 1, 2, 3, 4] },
    "heartrate": { "data": [140, 150, 152, 155, 158] },
    "latlng":    { "data": [[38.603734, -122.864112], "…"] },
    "altitude":  { "data": [10.2, 10.6, 11.1, 11.0, 10.8] }
  }
}

Output (selected records)

A heart-rate stream → a single-channel sampleArray Measurement:

{
  "id": "strava-14582931007-hr",
  "recordType": "Measurement",
  "subject": "subj-001",
  "type": "heart_rate",
  "unit": "/min",
  "sampleArray": { "offsets": [0, 1, 2, 3, 4], "dataPoints": [140, 150, 152, 155, 158] },
  "startTime": "2026-06-20T06:30:00Z",
  "endTime": "2026-06-20T07:13:20Z",
  "provenance": { "method": "sensor", "sourceApp": "strava", "device": { "manufacturer": "garmin", "model": "Garmin Forerunner 965" } }
}

The lat/lon/alt streams → one multi-channel route Measurement:

{
  "id": "strava-14582931007-route",
  "recordType": "Measurement",
  "subject": "subj-001",
  "type": "location",
  "sampleArray": {
    "offsets": [0, 1, 2, 3, 4],
    "channels": [{ "name": "lat", "unit": "deg" }, { "name": "lon", "unit": "deg" }, { "name": "alt", "unit": "m" }],
    "dataPoints": [[38.603734, -122.864112, 10.2], "…"]
  },
  "startTime": "2026-06-20T06:30:00Z",
  "endTime": "2026-06-20T07:13:20Z",
  "provenance": { "method": "sensor", "sourceApp": "strava", "device": { "manufacturer": "garmin", "model": "Garmin Forerunner 965" } }
}

The average HR → an interval quantity aggregate with derivedFrom:

{
  "id": "strava-14582931007-hr-mean",
  "recordType": "Measurement",
  "subject": "subj-001",
  "type": "heart_rate_mean",
  "quantity": 152,
  "unit": "/min",
  "startTime": "2026-06-20T06:30:00Z",
  "endTime": "2026-06-20T07:13:20Z",
  "links": [{ "type": "derivedFrom", "ref": "strava-14582931007-hr" }],
  "provenance": { "method": "algorithm", "sourceApp": "strava", "device": { "manufacturer": "garmin", "model": "Garmin Forerunner 965" }, "algorithm": { "name": "strava-summary", "version": "v3" } }
}

The activity → a Session whose continuous WorkUnit references the telemetry:

{
  "id": "strava-14582931007",
  "recordType": "Session",
  "subject": "subj-001",
  "disciplines": ["run"],
  "intent": "train",
  "startTime": "2026-06-20T06:30:00Z",
  "endTime": "2026-06-20T07:13:20Z",
  "workUnits": [
    {
      "id": "strava-14582931007-wu",
      "recordType": "WorkUnit",
      "scoring": "continuous",
      "performance": { "distance": { "absolute": { "value": 10000, "unit": "m" } }, "time": { "absolute": { "value": 2520, "unit": "s" } } },
      "links": [
        { "type": "measuredBy", "ref": "strava-14582931007-hr" },
        { "type": "measuredBy", "ref": "strava-14582931007-route" }
      ]
    }
  ]
}

derivedFrom ⇒ provenance.algorithm (recommended)

A derivedFrom link makes provenance.algorithm recommended (§7.4, relaxed in v0.3.1 from required). Strava doesn’t publish its aggregation algorithm, so the mapper supplies a best-effort algorithm.name ("strava-summary"). When you genuinely can’t name the algorithm, omit it — the relaxed rule exists precisely so a faithful mapper never has to drop the lineage link or fabricate a value.

Notes

• continuous scoring may carry any of distance, time, energy and requires no single metric — the endurance shape works end to end.
• endTime is start_date + elapsed_time (2600 s → 07:13:20Z); moving_time (2520 s) is the WorkUnit’s recorded time.
• peerSensor (a parallel co-stream, e.g. two HR straps) is distinct from sameActivityAs (whole-activity dedup across sources) — see the data model.