Migraine for AI

Migraine for AI — log each migraine attack in a single tap and let statistics surface possible triggers.

CORE
• One-tap logging — big button, current timestamp. Backfill any time.
• In-progress tracking — forgot to stop? An ongoing-migraine banner with a live timer waits on the dashboard until you tap End.
• Intensity, symptoms, triggers, location, medication, notes — all optional.

BACKGROUND CAPTURE (with your permission)
• HealthKit — heart rate, HRV, SpO2, respiratory rate, activity, sleep snapshots every ~4 hours.
• Weather — temperature, humidity, barometric pressure, UV, sun (Apple WeatherKit).
• Air quality (AQI / PM2.5 / O3 / NO2) — optional, requires your own OpenWeatherMap API key in Settings.

STATISTICAL ANALYSIS
• Compares ±N-hour windows around each attack against your background snapshots.
• Cohen's d effect sizes, 1000-iteration bootstrap 95% CIs, L2-regularized logistic regression weights.
• Reverse-causation factors (heart rate, HRV during an attack — they shift because of the migraine, not the other way round) are flagged honestly and excluded from trigger ranking by default.

REPORTS
• Self-contained Markdown export you can paste into your preferred AI assistant, or hand to your doctor.
• Event list, per-event detail, top factors, time-of-day / day-of-week distributions, plus a hand-off prompt that tells the LLM how to reason about your data.
• Fully offline — nothing is uploaded.

PRIVACY
• Records live on-device via SwiftData.
• HealthKit and location data are read and analyzed locally only.

THINGS TO KNOW
• Meaningful analysis needs samples: roughly 5+ logged attacks and 30+ background snapshots.
• Not a medical device. Persistent or worsening migraines need a clinician, not an app.