REF-397 · Altini M, 2024 · Gatorade Sports Science Institute, Sports Science Exchange, Vol. 37, No. 250, pp. 1-6

Making sense of wearables data

C+thin evidenceMeasured on a mixed groupWritten before October 2026, being re-checked

Key Results — 3-Tier Metric Hierarchy (checked against the full text 2026-09-29)

1

Measurements (directly captured by sensors): resting HR, HRV at rest, skin temperature — most reliable

2

Estimates of known parameters (derived via algorithms): calories burned, sleep stages, SpO2 — less reliable

3

Estimates of unknown parameters (composite scores): readiness, recovery, stress, strain — least reliable; no reference standard exists against which to validate them

Key Recommendations (as stated in the article; not verified against an abstract)

Establish a training plan first, then use wearables to capture responses — not the reverse
Focus on measurements, not estimates; focus on resting physiological data at standardised times
Treat composite scores that combine physiology and behaviour with scepticism — they lack reference standards
Use a "smallest worthwhile change" approach: compare today's value with the athlete's own historical range (full text, checked 2026-09-29: the article's example window is 30-60 days of data and a normal range of mean ± 1 SD)
Software updates alter calculations — longitudinal comparisons can become invalid

Mechanism

Error compounds across estimation layers: optical sensors lose accuracy during movement; composite scores combine already-estimated inputs with behavioural assumptions, producing an illusion of precision. No scientific reference standard exists for a "recovery score" — it is a proprietary construct.

Practical Application

Altini's hierarchy: trust what the sensor measures (resting HR, resting HRV, skin temperature) more than what an algorithm estimates (calories, sleep stages, SpO2), and trust composite readiness/recovery/strain scores least, because nothing exists to validate them against. Track your own resting trends over weeks, not a daily score.
Not from this paper; the base's own mapping: commercial "recovery", "body battery" and "readiness" scores fall in the third tier of this hierarchy; resting HR and HRV trends in the first
Do not make training decisions on a single-day composite score; compare a measurement with the athlete's own historical range (full text, checked 2026-09-29: 30-60 days, mean ± 1 SD)
Not from this paper: the measured nocebo of being told your sleep was bad is in REF-398 and REF-399 — a reason to hide daily scores, not just to distrust them

Limitations

Expert opinion, not a systematic review; no DOI and no abstract — the summary above is not verified
GSSI-affiliated (Gatorade funding)
Does not test its recommendations empirically
How this entry was checked and changed
Why this level

regraded 2026-09-27 from B+: narrative expert review without systematic method; no DOI exists (SSE No. 250), full text checked 2026-09-29

Content correction 2026-09-27
No DOI and no abstract exist for this article -> header says so; Level B+ -> C+ (narrative expert review, unverified).
"30-60 day historical data, mean ± 1 SD" -> kept only as a labelled full-text number; brand names (WHOOP, Garmin Body Battery, Oura) -> relabelled as the base's own mapping onto the hierarchy; "ignore daily recovery %; track 30-day trends" -> merged into the labelled line.
Cross-reference labels pointed at the wrong entry or named the wrong author (checked against each cited entry's byline); repointed to the entry the label describes, or removed where the base holds no such paper.
2026-09-29: the article named in the entry (SSE Vol. 37 No. 250) located and read on gssiweb.org (April 2024, Marco Altini; no DOI — the SSE series issues none). The full text confirms the three-tier hierarchy (measurements / estimates of known parameters / estimates of unknown parameters) and the normal range as mean ± 1 SD of the past 30-60 days; labels updated accordingly. Level unchanged (C+, expert review).