How to Track Cognitive Change over Time

A single score is not a trend

One number from one day answers almost nothing. A person who scores well on a Tuesday might score poorly on a Thursday because of sleep, a cold, or a noisy room. The number that tells you something is the direction across many days, and the only way to get it is to measure the same thing often, under roughly the same conditions.

So the goal is not a better test. It is a line. A line can be flat, drifting up, or drifting down, and each of those is a different conversation. A single reading is a dot that you cannot use.

Why appointments are too far apart to show change

Clinical cognitive assessment is usually repeated every one to three years, which is the right cadence for a diagnosis and the wrong one for watching a person week to week. Between appointments there is a long stretch in which a family notices something but has nothing to compare it against except memory and worry.

That gap is the reason home tracking exists. It does not replace the assessment; it fills the space between them with a light, regular signal. Done well, it makes the next appointment more useful, because you arrive with a direction of travel instead of a feeling.

What a usable baseline looks like

A baseline is a first week of the same activity, recorded, so later weeks have something to sit against. It should be a game baseline, not a clinical one: a rough picture of how this person performs on a good-ish day when they are not being tested. Three properties make it usable:

A baseline is not a benchmark against anyone else, and it is not a target to beat. It is this person, on an ordinary week, on this activity. Getting that framing right matters, because the moment a baseline is read as a score the behaviour changes: people try harder on a measured day, they get anxious when the number dips, and the line stops reflecting ordinary life.

Keep the conditions plain and repeatable — the same short session, the same time of day, the same few activities — and date every entry, so that months later a family can read the trend against time rather than guess from memory. Those habits are what make the three properties below worth protecting. A record kept casually is still far more useful than the impression you reconstruct in a waiting room.

• Same shape each time — the same few activities, so the comparison is honest.
• Low stakes — the point is a reference, so it must not feel like an exam the person can fail.
• Written down — a number you remember is not a baseline; a number you can look up on Monday is.

Reading a weekly report without over-reading it

A weekly report is useful precisely because it refuses to say too much. Sanospark's weekly family report sends four trend lines — speed, memory, attention and activity — plus one plain sentence and, at most, one flag. The design rule behind it is worth borrowing even if you never use the product: a report should make a trend readable, not turn a wobble into an alarm.

Three habits keep a report honest. Compare weeks to weeks, not days to days. Ask whether a change held for several sessions rather than one. And treat any flag as a question to raise, never a verdict — the report itself says as much.

Describing a trend to a doctor

A family who arrives saying "she seems worse" has little to offer. A family who arrives saying "over six weeks her reaction time in our home practice drifted down while her memory stayed flat, and she needed more prompts starting in the third week" has given the clinician something concrete. The second sentence is not a diagnosis and does not pretend to be; it is an observation with a shape.

That is the whole value of tracking well. It converts a vague worry into a short, dated observation, and leaves the interpreting where it belongs. The report cycle is built to produce exactly that sentence, and the framing it uses — worth mentioning at the next check-up — is the right level of confidence for a home signal.

What to keep, and for how long

Keep the raw weekly numbers, not a summary you rewrote later. A plain list of the same few metrics by date is enough; it is more honest than a remembered impression and easier to show than a screen full of charts.

Six to twelve weeks is usually enough to see a direction, and a direction is what you want. If you are tracking for a specific reason — a change a doctor is already watching, or a period after an illness — agree in advance how long you will gather and what you will do with it, so the record has a purpose rather than drifting into surveillance.

If the family's harder problem is that the record lives in one person's head, Easikin is an SI (artificial intelligence AI) caregiving app for adult children coordinating a parent's care, and Legaris is built around keeping the documents and plans that matter ready. For the training side of the loop, cognitive and physical training for older adults explains what the daily practice is doing, and memory and attention training explains why the memory and attention lines can move independently. Some care homes pair a home trend with their own activity record, and SilverGuard, the parent company, works on the residential-care sensing side of that picture. What counts as a session, and how long one should be, is set out in how long a brain training session should be.