OutPerform

A recovery / strain / sleep / readiness app for iOS that analyses your health data entirely on-device, using published sports-science methods (RMSSD/DFA-α1 HRV, Banister TRIMP, Gellish max-HR, Foster monotony, ACWR, Astrand VO₂, ODI).

Built with SwiftUI + SwiftData + Swift Charts, targeting iOS 26.5. No backend, no subscription — everything runs locally.

What it does

Feature	Where
Recovery score (HRV/RMSSD, RHR, resp rate, sleep, skin-temp & SpO₂ penalties, DFA-α1)	Algorithms/RecoveryCalculator.swift, HRVAnalysis.swift
Strain 0–21 (TRIMP, HR zones, muscular-load, ACWR, monotony)	Algorithms/StrainCalculator.swift
Sleep metrics + performance + apnea screening (ODI)	Algorithms/SleepCalculator.swift, SleepApneaScreener.swift
Sleep need (dynamic model), bedtime planner, chronotype	Algorithms/SleepNeedCalculator.swift
Stress monitor + breathwork (physiological sigh, box, cyclic)	Algorithms/StressCalculator.swift, Views/BreathworkView.swift
Health monitor (5 vitals vs personal baseline) + illness heuristic	Algorithms/HealthMonitor.swift
VO₂ max (HealthKit value or HR-ratio estimate)	Algorithms/VO2MaxCalculator.swift
Unified Readiness score, 7/30/90-day trends, daily check-in	Algorithms/ReadinessCalculator.swift, Views/TrendsView.swift

Items in bold are extra metrics beyond the basics.

Data sources (both supported)

The app reads through a single HealthDataProvider protocol (Health/HealthDataProvider.swift), so the source is pluggable. Pick one in onboarding or Settings:

• Apple Health — HealthKitProvider (the default). Reads HR, HRV (SDNN), RHR, respiratory rate, SpO₂, sleeping wrist temperature, sleep stages, workouts and VO₂ max directly from HealthKit. For HRV it reads the beat-to-beat heartbeat series (HKSeriesType.heartbeat() + HKHeartbeatSeriesQuery) when an Apple Watch recorded it, giving a true RMSSD and DFA-α1; otherwise it falls back to the SDNN value your wearable wrote, so you always get a recovery score. Requires the HealthKit capability (and a device with real Health data).
• Google Health — GoogleHealthProvider. Talks to the Google Health API (health.googleapis.com/v4, the platform replacing the Fitbit Web API). Auth is Google OAuth 2.0 (PKCE via ASWebAuthenticationSession, access_type=offline); data is read from the dataPoints REST endpoints. Scopes: googlehealth.health_metrics_and_measurements.readonly, …activity_and_fitness.readonly, …sleep.readonly, …profile.readonly. Data types: heart-rate, daily-heart-rate-variability, daily-resting-heart-rate, respiratory-rate, oxygen-saturation, daily-sleep-temperature-derivations, sleep, vo2-max, exercise. HRV is the daily summary (RMSSD/DFA need a beat-to-beat series, which Google doesn't expose), so recovery uses it the way it uses HealthKit's SDNN.

Configuring Google Health

1. In Google Cloud Console, create a project, enable the Google Health API, and create an OAuth 2.0 iOS client.
2. Paste the client id into Health/GoogleHealth.swift → GoogleHealthConfig.clientID. The OAuth redirect (the reversed-client-id scheme) is derived automatically and intercepted by ASWebAuthenticationSession, so no Info.plist change is needed.
3. Until a real client id is set, the Google option reports "isn't set up yet" and the app stays on Apple Health.

Note: the dataPoints value field names aren't fully specified in the migration docs, so the JSON parser is deliberately schema-resilient (it digs out the first numeric/stage value). Verify against a live account and tighten if needed.

Architecture

Models/        CoreTypes (samples, stages, zones), DailyRecord (SwiftData), AppSettings
Algorithms/    Pure, dependency-free score maths (unit-testable with `swiftc`)
Health/        Provider protocol + HealthKit / Google Health backends + Keychain
Engine/        ScoreEngine (orchestration), Background & Notification managers
Views/         SwiftUI screens (Today, Sleep, Strain, Body, Trends, Settings, Onboarding)
App/           AppModel coordinator

ScoreEngine.refresh() bulk-fetches the range once, then walks days chronologically so baselines, sleep debt and the ACWR EWMA carry forward correctly, persisting one DailyRecord per day.

Building & running

open OutPerform/OutPerform.xcodeproj    # then ⌘R, or:
xcodebuild -scheme OutPerform -destination 'platform=iOS Simulator,name=iPhone 17' build

The app reads only real data, so run it on a device with Apple Health data (or a connected Google Health account). You must enable the HealthKit capability for the App ID (automatic signing usually handles this via OutPerform.entitlements).

Calibration (spec §10)

Days 1–7 collect data, 8–13 show provisional scores, day 14+ full scores once baselines are reliable, day 30+ unlocks personalised weight learning.

Disclaimer

Not a medical device. The sleep-apnea (ODI) and illness features are informational screening only and never a diagnosis.