Performance Dossier — Simulated Telemetry
Three consenting operator-days, read from wearable telemetry and returned as curated, cited advisories. The raw numbers live in the operator’s own ring; the judgment — what they mean, and which vetted protocol answers them — is the instrument OCCABUZZ delivers. Select an operator to inspect the call.
Select operator
The Red-Eye Founder — closed a round in London, flew home overnight, back-to-back today.
These findings are not independent. Clock displacement is upstream: when timing shifts, sleep shortens, and short sleep is what drags autonomic recovery and raises overnight temperature. Treating each signal separately spreads effort across symptoms. Realign the clock and the others follow it back.
Four links carry a signal from skin to advice. Two are validated in peer-reviewed literature. Two are not, and we name them rather than let the polish imply otherwise.
Accuracy 91.7–91.8%, sensitivity 94.4–94.5%, specificity 73.0–74.6% against multi-night ambulatory polysomnography (n=96, 421,045 epochs). No significant difference from PSG for time in bed, total sleep time, onset latency, WASO, light or deep sleep.
Svensson et al., Sleep Medicine 2024 · doi:10.1016/j.sleep.2024.01.020
Staging accuracy 75.5% (light) to 90.6% (REM); independent replication puts stage sensitivity at 76.0–79.5%. In a six-device comparison, multi-state agreement was 61% (κ=0.43) — the best of the six, which is a statement about the field as much as the device.
Robbins et al., Sensors 2024 · doi:10.3390/s24206532 — Miller et al., Sensors 2022 · doi:10.3390/s22166317
HRV balance, sleep balance, recovery index and activity balance are Ōura's own constructs. No peer-reviewed validation exists for how they are computed or what a given value means. This is the weakest link in the chain, and it is not ours to strengthen.
The numeric thresholds are engineering judgment about where a contributor stops being unremarkable. They are not clinical cutoffs and no study supports the specific numbers. They were chosen to surface signal without crying wolf, and they will move when real operator data says they should.
Read as directional context, never as a medical determination. Nothing here diagnoses, treats, or replaces a clinician.
This is a curated performance readout, not medical advice, not a diagnosis, and not a prescription. Signals are directional context from a consumer wearable. The operator owns the data; every read is generated from a grant that can be revoked or deleted at any time. If unwell, consult a licensed clinician.
OCCABUZZ// · Simulated Trust Loop (Phase 0.5) · Synthetic Oura v2 Data · No PII