Sources
| Source | Provides | Category |
|---|---|---|
| Garmin device + Garmin Connect | Distance, duration, pace, heart rate, elevation, cadence; algorithm-estimated VO₂ max, resting heart rate, sleep staging | Device measurement / device estimate |
| Home blood pressure monitor | Blood pressure (validated, upper-arm cuff) | Home reading |
| Bathroom scale (consistent unit & conditions) | Weight | Home reading |
| Body-composition method (method to be confirmed at first checkpoint) | Body fat %, muscle mass | Home reading or clinical, depending on method chosen |
| Laboratory panel | Selected blood values (glucose, lipids, and others as relevant) | Laboratory |
| Logged strength protocol | Strength benchmark set | Manual, logged |
| Self-report | Perceived effort, pain, fueling notes, sleep quality | Subjective report |
Cadence
- Training data — every session, from Garmin.
- Resting heart rate & sleep — nightly from Garmin, read as a rolling weekly average rather than single-night swings.
- Body composition & strength benchmarks — roughly monthly checkpoints, same method each time where possible.
- Blood pressure & glucose — baseline, a mid-project checkpoint, and post-race.
- Full blood panel — baseline and shortly after the marathon.
Estimate vs. measurement
Garmin's VO₂ max is a Firstbeat-algorithm estimate derived from pace and heart rate during runs — useful for watching a trend over months, not equivalent to a laboratory gas-exchange test. Where both exist for the same metric, they're kept as clearly separate series with a visible note, never merged into one line. See the VO₂ max explainer for the full version of this distinction.
Limitations
- One person, not a controlled study — see confounders & limitations on the Experiment page.
- Home readings vary more than clinical ones with technique, time of day, and conditions; conditions are recorded alongside each reading where they matter.
- A change in device, cuff, or scale is flagged as a method change, not folded silently into the existing series.
- Sparse data (fewer than a few comparable readings) is marked Insufficient Data rather than charted as a trend.
What's not published
Original lab and clinical reports stay private. The public site carries selected values, units, dates, methods, and interpretation — not scanned documents, and not identifiers unrelated to the measurement itself.