Swift 6 · HealthKit · SwiftUI · WidgetKit · iCloud
Current gives you a daily 0–100 recovery score built from your own Apple Health data.
I designed the scoring algorithm from scratch: HRV at 45%, sleep at 35%, resting heart
rate at 20%, each scored against a personal baseline rather than a population average.
The weights were validated against 30 days of real HealthKit data, then calibrated
against a clinical polysomnogram.
Sleep stages are intentionally excluded from the score. Wearable sleep staging is
unreliable and device-specific. An in-lab study confirmed Apple Watch reported 29
minutes of deep sleep against 70 minutes measured clinically. HRV already captures
the recovery outcome that quality deep sleep produces, so scoring stages would reward
hardware, not recovery.
The engineering is production-grade throughout. Swift 6 strict concurrency, HealthKit
background delivery, iCloud sync, four widget sizes, and a custom guidance engine that
reads the full health snapshot so the message it shows can never contradict the
sub-scores. I kept architecture decision records throughout development and maintained
a gotchas log for every concurrency edge case. The test suite covers 82 unit tests
across the scoring engine, trend detection, notification rules, and insight generation.
Built with Claude Code
Building a production iOS app solo means owning every decision: the scoring math,
the HealthKit concurrency model, the data architecture, the test strategy. I brought
the domain knowledge and made every call. Claude Code was the development partner
throughout, reading the full codebase before touching anything, enforcing the Swift 6
constraints I documented, and running tests after changes. The expertise is mine.
Claude Code made sure nothing slipped through.
How it was built →