FeaturePatterns

It tells you how sure it is.

After a few weeks your log knows things you don’t. Patterns reads it and says what stands out, and every finding says what it is: something that happened, an early sign, or a pattern that has earned the word.

Download See how it works
An illustration in soft white clay: three cards standing one behind the other, more strongly tinted from back to front, three columns rising beside them with the tallest one glowing raspberry, and a small round badge with a tick.
Graded, every time
Observation, early sign or pattern
30 days for a pattern
And a false-discovery check across everything
Associations, not causes
Said on the page, not in small print
Free
Highlights and their grades

01Highlights

What stands out in this range

Five cards, not fifty charts.

Open Patterns and a drawer of Highlights rises over the body: at most five cards, the strongest first. Each is one sentence in plain words, with its grade above it and its evidence under it.

Longnote’s Patterns tab in a loop: the Highlights drawer rises from its resting height, its cards scroll past, one opens the correlation explorer, and the drawer settles back down.
  • 01Patterns by the hour first: how a symptom moves in the hours after a dose, “Joint stiffness was about 23% milder 1–4 h after you took Ibuprofen”, once there are enough doses to say so.
  • 02Then patterns between days: two trackers that move together, on the same day or up to three days apart.
  • 03At most one early sign at a time, so a hunch never crowds out what is known.
  • 04Then observations about a tracker on its own: “On 18 of 30 logged days”, or its average against the span before.

02How sure

Four standings, weakest first

A pattern has to earn the word.

Compare enough things and some will line up by chance. So the app counts how many comparisons it made, and only a link that survives that check across all of them is called a pattern.

  1. Observation
    Any amount of data

    A fact about one tracker’s own entries: “On 18 of 30 logged days”. It cannot be wrong, and it claims nothing about why.

  2. Hint
    3 or 4 days on one side · lists only

    A possible link where one side of the comparison has only 3 or 4 days. Shown only in browsing lists, like the weather sheet and Moves with it, with both counts. Never in Highlights or the report.

  3. Early sign
    Not yet confirmed · in Highlights, under 60 days of overlap

    A possible link between two things, worded as a hunch with its day count. Same day or next day only. Highlights show at most one at a time.

  4. Pattern
    At least 30 days of overlap · 1-in-10 false-discovery rate

    A link that survived a false-discovery check across everything the app compared, and still points the same way in the range on screen.

How strong

A pattern says how closely the two move: a weak, moderate or strong link, with the days behind it.

How much evidence

Limited, fair or good evidence, from the number of days and whether the link holds when the timing shifts. Limited evidence is never called strong.

Still true now

A pattern must still point the same way in the range on screen, or it is not shown.

03What it reads

Your entries, and nothing else

Everything you log is a question it can ask.

There is no list of suspects to fill in. Every tracker is compared with every other, and the things you mark become factors of their own.

Trackers with trackers

Sleep and fatigue, coffee and migraine, a treatment and the symptom it is for: the same day, or one to three days later.

The options you pick

Each option of a choice measure, like a food or a trigger you tick, is checked as its own factor once there are enough days with and without it.

Pressure drops and rain

With Pro weather, days when the air pressure fell 3 hPa or more, and rainy days, are compared with the rest.

Everything is computed on your iPhone from your own entries. They are never sent anywhere to be analysed.

04Look closer

Pro

See exactly what a finding stands on.

With Pro, tap a Highlight and it opens with what it is based on: the two averages, with and without, the days on each side, and how the link changes when one is shifted a day or more.

  • 01Compare two trackers on one timeline.
  • 02Correlations: how any two move together, the same day or up to three days apart.
  • 03Heatmap: months of one tracker at a glance.
  • 04What moves it: the factors most linked to one tracker, ranked.

Your say

Each finding you open ends by asking whether it rings true for you. You know things your log doesn’t.

Longnote’s Correlations sheet, Migraine against Overall mood, same day: “Overall mood was worse than usual on 100% of days your migraine was Mild or worse. On other days: 55%”, a strong link. Below it, What this is based on: 29 days compared with 136 other days, 165 days with both logged, average overall mood 1.7 against 3.4, and a small chart of the link at the same day and one to three days later.

05Honest by design

What it will not say

Never a cause. Never advice.

A link in your log is a reason to look closer, or to bring it to an appointment. It is not proof, and the app says so where you read it.

Associations

Under the Highlights, every time: Patterns are associations, not causes.

Early signs fade

When one is shown, the page adds that early signs need more days and often fade. Many do.

Nothing to show

With too little data, Patterns says No patterns yet rather than inventing one.

06Read next

Blog · Guides

More on reading your own data.

Blog How Highlights decide what to say Observations, early signs and patterns, and the numbers behind each. Guide How long to track before a pattern means anything Why a few weeks is the floor, and what changes after three months. Guide Is a trigger the same as a cause? What a link in your log can and cannot tell you.

07Explore more

Features

The rest of the note.

Let the weeks add up.

Free to download, free to keep. Highlights and their grades are free, and the analysis never leaves your iPhone.

Download Longnote All features