How the pattern view is worked out

Counting, not prediction. Every line in the app shows the counts it's based on.

What gets compared

An attack day is a day you logged an attack starting. An attack-free day is a day you answered "No" in the daily check-in. Days with no answer are left out.

When it unlocks

Emerging patterns after 4 logged attacks; Established patterns after 8 attacks and 20 check-ins.

The rule for each factor

We count attack days with the factor out of attack days where it was recorded, and the same for attack-free days. A factor is listed as "shows up often on your attack days" only if: at least 4 attack days and 5 attack-free days have it recorded; it appears on at least 3 attack days; the attack-day share is at least 20 percentage points higher; and a one-sided Fisher's exact test gives p ≤ 0.1. The reverse rule gives "shows up often on your good days". Everything else is "no clear pattern" or "too little data".

Factor definitions

• Air pressure dropped sharply: a fall of 6 hPa or more (adjustable) within 24 hours; for an attack, within the 24 hours before it started. Data: Apple Weather hourly pressure for your rounded area, with Open-Meteo as a backup source.
• Hot day: high of 30°C (86°F) or more. Humid day: average humidity 80% or more.
• Short sleep: under 6 hours asleep the night before (Apple Health).
• Period window: 2 days before to 3 days after the first day of a period you marked.
• Other factors: what you tapped or said. A factor counts as absent only on days you checked in.

Patterns are counts from your own logs, shown with sample sizes. Correlation in your logs is not proof of cause. A diary supports your clinician; it never replaces them.