Compute the sleep robust-z scales once per night, not per epoch

[abdulsaheel]

cardioStager scored three fixed samples (sleepRmssd, sleepLfhf,
sleepRk) with robustZ inside the per-epoch classify loop. Each call
re-derived the same median and MAD, sorting the sample twice — about 2,880
times over a night, on lists that never change inside the loop.

Hoisted to a RobustScale built once after the samples are assembled.
robustZ now delegates to it, so the null semantics are shared rather than
duplicated: absent for a sample under two values or a MAD of zero.

Output is unchanged. Verified by running both versions over an 8-hour
synthetic night and comparing the stage sequence, deep flags and confidence —
identical. 765 ms -> 408 ms per run.

Tests: 383 pass.

Summary by CodeRabbit

• Performance
  • Improved sleep-stage analysis efficiency by reusing statistical calculations across sleep epochs.
  • Reduced repeated processing during REM-related detection without changing results or public interfaces.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 8d42cb08-0461-4da0-91c5-802a0903f15e

📥 Commits

Reviewing files that changed from the base of the PR and between f5ccae6 and 832d8fe.

📒 Files selected for processing (2)

• lib/src/onehz/sleep/cardio_stager.dart
• lib/src/onehz/util.dart

---

📝 Walkthrough

Walkthrough

RobustScale precomputes median and MAD-derived values for reusable robust z-score calculations. cardioStager creates cached scales for sleep baselines and uses them during REM autonomic classification.

Changes

Cached Robust Scaling

Layer / File(s)	Summary
RobustScale abstraction lib/src/onehz/util.dart	Adds RobustScale and updates robustZ to use precomputed median and MAD-derived scale values.
Cardio staging scale reuse lib/src/onehz/sleep/cardio_stager.dart	Precomputes RMSSD, LF/HF, and R(k) scales once per staging run and reuses them in REM gating calculations.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• OpenStrap/analytics#25: Modifies the same REM autonomic robust-z feature calculations.
• OpenStrap/analytics#26: Handles degenerate robustZ results in related readiness scoring logic.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main change: caching sleep robust-z scales once per night instead of recomputing them each epoch.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/optimizations

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}