What goes on a line
Five columns separated by a vertical bar: date | symptom | severity | time | what was going on. The first three are required and the last two can be left empty. Severity is 0 to 10, with 0 as nothing and 10 as unbearable. The fifth column takes short phrases separated by commas, and those phrases become the rows of the tally further down, so it pays to write the same thing the same way each time — poor sleep and didn't sleep will be counted as two separate things. Tabs work as separators as well as bars, so a table kept in a spreadsheet pastes across. If you are noting a temperature, put it in that last column and keep it in °F throughout rather than switching between scales.
A number you invent yourself
There is no instrument for severity. It is a number you assign, which means your 6 and somebody else's 6 are not comparable, and even your own 6 drifts over weeks. It is still worth recording, because the point is not the absolute value but the movement: a fortnight averaging 6 followed by a fortnight averaging 4 is a real change regardless of what the numbers mean to anyone else. Anchoring the scale helps it hold still. Decide once that "had to stop what I was doing" is a 7 and "noticeable but I carried on" is around a 4, then keep using those anchors. Whatever consistency you can manage in the scoring is the thing that makes the mean readable later.
Reading the tally without over-reading it
Each phrase from the fifth column gets a count and the mean severity of the lines it appeared on. It is co-occurrence arithmetic and nothing more. Three things make it unsafe to read as cause. What people record is shaped by what they already suspect, so the counts arrive pre-tilted. Symptom-free days produce no rows at all, so the control group is entirely missing — nowhere in this data is the coffee you drank on a day when nothing happened. And the entries cluster: late work, poor sleep and stress mostly turn up on the same lines, and no amount of averaging can pull them apart. What the table is genuinely useful for is choosing which two or three things to raise with someone who can test them properly.
The paragraph, and what it leaves out
The generated sentence packs the counts, the mean, the worst entry and the most common notes into something you can read aloud in fifteen seconds, which is often the difference between getting to the real question and spending the appointment on background. It contains only what the diary contains, and there are four things it structurally cannot know: when this started, whether it has happened before, what makes it better or worse, and what else you are taking. Those go in by hand — the printed sheet has an empty notes column for exactly that. One thing that does not belong in any diary: symptoms that arrive suddenly, feel unlike anything before, or come with weakness, confusion, chest pain or trouble speaking. Those want a phone call while they are happening, not a row added to a table for later.
Questions people ask
The tally shows a high severity next to one phrase. Is that the cause?
No, and the table cannot answer that question. It counts how often a phrase sat on the same line as a symptom and averages those severities. Days when nothing happened produce no entries at all, so the comparison is missing half the picture, and the phrases people write tend to arrive in groups that cannot be separated arithmetically. Use it to pick what to raise, and let the appointment sort out what matters.
How should I score severity?
However you like, as long as you keep doing it the same way. Fixing a couple of anchors first — say, 7 and above means stopping what you were doing, around 4 means noticeable but manageable — makes today’s score comparable to one from three weeks ago, which is the only comparison that matters here.
Is the diary stored?
No. It is all worked out in your browser, nothing is transmitted, and refreshing the page empties the box. Copy the text output somewhere safe if you want to keep adding to it; pasting it back reads in unchanged.