Human Fatigue & Error
Tracking fast-moving cells under strict laboratory time limits makes the daily process highly prone to eye fatigue and microscopic misclassifications. This is not a question of who is at the scope. It is what happens when sustained visual attention is the only instrument in the room.
Human Fatigue & Error explained
01
Work that cannot wait
Almost every judgement in an IVF lab sits inside a window that biology opened and will close on its own schedule. Oocytes are assessed and injected within a defined period after retrieval. Fertilisation is checked at a set hour. Blastocysts are graded in the interval when the structure is actually legible. Sperm are selected during the procedure itself, with the needle already loaded.
None of those clocks can be paused for a second opinion, a break, or a quieter afternoon. The assessment happens when it happens, at whatever point in the shift it lands, with however many dishes are still queued behind it.
02
What the eye is being asked to do
The visual task itself is at the edge of what human vision resolves. Sperm selection during ICSI means judging motion — track, speed, progression — through high-magnification motion blur, on cells that cross the field in a fraction of a second, and choosing one. Embryo grading means resolving sub-micron differences in layer boundaries across a translucent, three-dimensional specimen flattened onto one plane of focus.
Both are performed at maximum sustained attention, repeatedly, for hours. Visual acuity, contrast sensitivity and fine discrimination all fall away under that load — this is ordinary human physiology, not negligence, and no amount of experience switches it off. An embryologist ten years in is not immune to it. They are simply better at recognising when it has started.
The lab is not short of skill. It is short of a second observer who is as fresh at the end of the list as at the start of it.
03
Why nothing catches it
A misclassification made under fatigue leaves no trace. The output of the assessment is a written grade or a selected cell; the reasoning behind it exists only in the moment it was made. There is no image to revisit, no measurement to re-check, no record of which structures were weighed and how. If the reading was off, the lab finds out — if it finds out at all — from an outcome weeks later, by which point the cause is unrecoverable.
That is what makes this different from most quality problems in medicine. It is not that errors go uncorrected; it is that they go undetected, and therefore unmeasured. A lab cannot improve a step it cannot see, and it cannot staff around a failure mode it has no way to quantify.
The usual mitigations run out quickly. Double-reading costs a second embryologist per specimen, which the caseload will not absorb. Shortening lists reduces the cycles a clinic can run. Both trade throughput for confidence, and neither produces a record.
It is not that these errors go uncorrected. It is that they go undetected — and a lab cannot improve a step it has no way to see.
04
What has to change
The fatigue is not removable — the work is what it is. What can be removed is the lab’s dependence on sustained human attention as its only measuring instrument.
Forlivf runs on the microscope camera the lab already owns and watches the same field the embryologist is watching. It tracks every sperm in frame rather than the two or three a person can follow. It maps the structures of every blastocyst to the same standard at 8am and at 6pm. It does not get tired, and — the part that matters most here — it leaves a record, so the reading behind a decision still exists after the decision has been made.
The embryologist remains the one who decides. What changes is that they are no longer the only thing in the room that saw the specimen.
What answers it
Automated Gamete Precision
Computer vision calculates linear velocity vectors through high-magnification motion blur, isolating optimal cells for ICSI in milliseconds — the part of the work that most punishes a tired observer.
Explainable Embryo Grading
A structural read of every specimen that does not degrade over a shift, so the last assessment of the day is made against the same standard as the first.