Illness breaks all three of the app's inputs at the same time: you eat differently, you weigh differently, and you are in no state to log carefully. None of it does lasting damage if you handle the week deliberately.
Log what you can, mark what you can't
The rule from how your expenditure is calculated applies with more force than usual: a rough entry beats a blank, and a blank is the only thing that actually hurts.
If you managed a soup and some toast, log them roughly. If a day is genuinely unreconstructable, mark it incomplete logging — that removes it from the calculation entirely, which is much better than letting a half-logged day pass as a 700-calorie one. Fasting days, incomplete days and manual totals shows where the switch lives.
If the whole week went unlogged, that is what the Logging break module at your check-in is for. Your estimate holds steady and picks up when the data returns.
Ignore the scale for a bit
A few kilos can leave and come back inside a week of illness, and almost none of it is fat: less food in transit, glycogen and the water bound to it, sweat and fluids. Weigh in if you feel like it — the trend weight is smoothed precisely so that one strange reading cannot steer anything, and a weigh-in's influence halves in about eight days. What you should not do is read the daily number as progress in either direction.
If you stop weighing in altogether for more than a week, the estimate stops updating and holds its last good value. That is the system protecting itself, not losing your data.
Don't lock in a sick week's targets
The estimate may dip on the way out. Lower intake against a weight that dropped for reasons that have nothing to do with fat can read as a lower expenditure for a few days, and your next check-in will suggest targets to match.
You are not obliged to take them. The check-in always offers keep alongside apply, and a week you were ill is the textbook case for keeping — see apply or keep your targets. Give it a normal week of eating and weighing, and the estimate finds its way back on its own.
When it was more than a week
For a long illness, treat the return like a restart rather than a resumption: log completely for a fortnight before judging any number, and expect the estimate to spend two to three weeks catching up as the windows refill with normal data.