The first five thousand

What people say it has changed

The most common report we receive is not about a feature. It is about being believed. People describe showing the app to a partner, a parent or a doctor and finally having a picture of a day that matches how that day felt. For an illness routinely met with scepticism, an external record appears to carry weight that testimony alone does not.

The second most common report is more practical: people tell us they can now see the cost of a specific activity before they commit to it, and are declining things they would previously have attempted. Several describe this as the first time pacing has felt possible rather than merely instructed.

A third theme is subtler and, from a research point of view, the most interesting: people report that the app's assessment of their day frequently agrees with their own, and that when it disagrees, it is often early. Several users have described being warned on a day that felt acceptable, ignoring it, and paying for it two days later. We treat these accounts as anecdote rather than evidence, but they are consistent enough to be worth formal study.

One milestone worth recording alongside the feedback: FatigueSense is now available in five languages, English, Dutch, French, German and Spanish. These conditions are not confined to English-speaking countries, and neither is the shortage of tools for them. Further languages will follow.

What we are seeing in the data

What follows is description rather than conclusion. These are patterns observed across a large group of people using ordinary consumer wearables in daily life, not clinical findings, and we set them out because they have already changed how our software behaves and because they may be useful to anyone thinking carefully about this illness.

The first and most consistent observation is that the morning is the most expensive part of the day. Rising, washing, dressing and eating routinely account for a substantial share of a person's entire daily capacity, spent before nine o'clock and before anything most people would describe as activity. Users regularly assume this must be a fault in the software. It is not, and the implication is uncomfortable: advice that treats the morning as neutral preparation for the day is mispricing the single most expensive thing many patients do.

A related and under-appreciated point is that a substantial minority do not live on a calendar day at all. Many are awake through the small hours and asleep through the middle of the day. Any measurement anchored to midnight splits their worst hours across two records and hides the pattern inside both. This carries a direct implication for study design: analyses that bucket wearable data by calendar day are quietly mismeasuring precisely the group least able to tolerate being misunderstood.

Stillness is not rest. Lying motionless with an elevated, unsettled heart rate is not recovery, yet people in that state are routinely recorded by consumer devices as having had a restful day. Distinguishing “did not move” from “was at rest” matters enormously in this population, and almost no consumer device attempts it. The same blind spot explains why effort without movement goes unrecorded: among the most severely affected, days with essentially no movement can still carry substantial physiological work, as broken sleep, pain, distress, digestion and being cared for all register as genuine effort. Any framework equating exertion with voluntary activity misses this entirely, and misses it in the group with the least margin for error.

We can also say with some confidence that the same activity does not have a stable cost. An identical walk, taken by the same person a few days apart, can cost very differently depending on sleep, infection, heat, hormonal cycle and what preceded it. Fixed activity-cost tables, of the kind that appear in most written pacing advice, are wrong within a fortnight for most individuals, which may explain some of the gap between how sensible that advice sounds and how difficult people find it to apply.

On recovery, a pattern reported often enough to be worth stating plainly: a single good morning is a poor guide to whether an episode has ended. People describe waking part-way through a difficult period feeling, on that particular morning, deceptively normal, and then remaining unwell for days afterwards. Several have independently adopted a personal rule of waiting for two or three settled mornings before considering an episode over. It is a sensible habit, and it suggests that any judgement resting on one morning deserves caution.

Finally, something we did not anticipate but which has shaped the product more than any other single observation: people in this population ask what an activity will cost, not how they performed. Feature requests here are strikingly consistent and strikingly unlike those in general fitness. Nobody asks for streaks, badges or competition. They ask, before the event rather than after it, a single question: if I do this, what will it take from me, and will I still be able to afford Thursday?

What this suggests for practice and for research

Pacing is among the few interventions with real support in ME/CFS care, and yet it is typically prescribed without any instrument. Patients are asked to stay within a limit they have never measured, judged by how they feel, in an illness whose defining symptom is delayed. Feeling is a lagging indicator here, which makes unaided pacing structurally difficult in a way that is rarely acknowledged in the advice itself.

Three things follow from what we are seeing. Thresholds have to be individual, because population and age-based formulas are close to meaningless in a dysregulated autonomic system. Measurement has to work for people who do not move, because the most severely affected are the least well served by movement-based metrics. And recovery has to be defined over several days rather than a single morning.