Skip to content

ROADMAP.md

Latest commit

 

History

History
42 lines (30 loc) · 2.27 KB

ROADMAP.md

File metadata and controls

42 lines (30 loc) · 2.27 KB
Raw

Roadmap

Health Dashboard is a self-hosted, evidence-oriented health analytics system. The roadmap is intentionally conservative: health signals should become more useful only when their inputs, confidence, and limitations are visible.

This document is directional, not a release promise. GitHub Issues are the source of truth for scoped work.

Now

  • Finish EnergyBank v2 operational hardening: deployment verification, stress observability, and post-rollout cleanup.
  • Add focused tests around current scoring and import behavior.
  • Keep documentation aligned with shipped methodology so contributors can audit what each score means.
  • Improve OSS maintainer workflow: issue templates, PR templates, labels, and contributor guidance.

Next

  • Define a readiness freshness and confidence contract that the dashboard can render without guessing.
  • Add confidence and uncertainty indicators for user-facing scores.
  • Validate readiness, EnergyBank, and stress signals against subjective check-ins once enough answered days exist.
  • Expand anonymized fixtures for Apple Health parsing and import safety.
  • Review authentication/session hardening for non-local deployments.

Later

  • Revisit parked readiness experiments only after serving/freshness and monitoring gates are operational.
  • Explore workout-derived readiness only after structured workout import exists at meaningful scale.
  • Decide whether to add slower CI jobs such as the race detector.
  • Consider structured logging only when there is a concrete operational need.

Explicitly Not Planned

  • Medical diagnosis or clinical decision support.
  • Fabricating wellness scores when required inputs are missing or stale.
  • Uploading raw personal health history to a hosted analytics backend by default.
  • Shipping scoring changes from private-data experiments without documenting the evidence boundary.
  • Treating population benchmarks as a replacement for personal baselines.

Working Principles

  • Raw data stays preserved and derived layers stay rebuildable.
  • Missing, imputed, stale, and low-confidence states should be visible.
  • Scoring and calibration changes need tests, methodology notes, and a clear rollback path.
  • Public issues should use anonymized language such as Profile A / Profile B, not real tenant names or personal health values.