Accepted answer
Week 3 is day 21: on a four-week ladder that is week 3 of dose step 1, and — at the seven-day half-life this class runs on — 3 half-lives in. Steady state is about five half-lives, so day 35 is where the concentration stops climbing on its own. Day 21 is 2 weeks short of it, so the level is still rising even though the dose has not changed. That distinction is most of the question: at week 3 of a step, an effect that is still accumulating is indistinguishable from one that is not resolving unless you know which side of day 35 you are on. Local erythema is a site-and-technique observation before it is a drug observation. It should track where and how you injected rather than how many weeks you have been injecting, so the useful record is a rotation log rather than a week count. Dose decisions are made under supervision, and nothing here is medical advice.
Photographing the site with a date is worth more than describing it later from memory.
Features that point towards infection rather than reaction: expanding redness beyond a few centimetres, increasing rather than decreasing pain after forty-eight hours, warmth, fluctuance, fever or malaise.
Marking the edge of an area of redness with a pen and a time is the cheapest way to establish whether it is expanding, and it converts an impression into an observation.
Injection-site reactions are reported in the trial programmes for this class at low single-figure to low double-figure percentages depending on agent and formulation.
Mark the edge with a pen and note the time. Expanding is the signal that matters.