A Mixed Reality Surgery Trainer That Changed Nothing
A pilot study of mixed reality training for vascular anastomosis found no significant difference between groups, in either session. We are covering it because of that.
An exploratory pilot study of mixed-reality-assisted training for vascular anastomosis found no significant difference between groups. First-session scores were 28.3 and 32.0 (p = 0.106); second-session scores were 31.8 and 30.7 (p = 0.467). The groups did not pull apart by the second session, which is what you would expect if the training had stuck.
What happened
Joining two blood vessels is called an anastomosis. It is delicate hand work, and trainees need many repetitions to get good at it.
Mixed reality is an obvious candidate to help. The trainee wears a headset and sees guidance laid over their hands and the vessel. Where to place the stitch. How far apart. Dozens of studies in medicine have reported that this kind of guidance helps.
A team ran an exploratory pilot study to see whether it helped here. An exploratory pilot means a first, small look, run to find out whether a bigger study is worth doing. It is not designed to settle anything.
The test
Two groups were compared, and the trainees were scored twice — a first session and a second session some time later. Testing twice matters, because it separates “did better on the day” from “learned something that stuck”.
The result
Nothing moved.
In the first session the two groups scored 28.3 plus or minus 3.1, and 32.0 plus or minus 2.5. The p value was 0.106.
In the second session they scored 31.8 plus or minus 2.8, and 30.7 plus or minus 1.7. The p value was 0.467.
A p value is roughly the chance of seeing a difference this big if there were really no difference at all. Researchers usually treat 0.05 as the line. Both of these numbers are well above it. So in plain language: this study did not find that the mixed reality training worked.
Note what also did not happen. The groups did not pull apart in the second session. If the headset were teaching something that stuck, you would expect the gap to grow by the second test. It did not. If anything the two groups drew closer together.
Why report this at all? This is because studies that find nothing mostly do not get written up, and the ones that do get written up rarely get read. That habit has a name: publication bias. It means the research you see about any technology is a filtered sample, weighted towards the times it worked. A NULL RESULT IS A RESULT.
What it means
Do not read this as “mixed reality does not work for surgical training”. One small pilot cannot show that, and this one does not claim to.
Read it instead as a correction to the weight of the evidence. If you had read ten positive studies about headset training and drawn a conclusion, this is the eleventh study, and it belongs in your average. Leaving it out is how a field talks itself into something.
For anyone deciding whether to spend money on this, the practical reading is: the benefit is not so large or so reliable that a small study picks it up automatically. Therefore, ask any vendor for the studies that found nothing. If they have never seen one, they have not looked.
Business ideas from this paper
1. A null-results digest for one profession. Somebody has to collect the studies that found nothing and put them where buyers will see them. Pick one field — surgical training, say — and publish the quiet half of the evidence. Buyer: training colleges, professional associations and hospital procurement. First test: send one issue to twenty training directors and count the replies.
2. Pilot-study design for clinics about to buy equipment. This study’s weakness is its size, and most hospitals run even weaker internal pilots. Sell the design: how many trainees, which score, how many sessions, what counts as success — agreed before any money is spent. Buyer: hospital education and training departments. First test: offer one design free in exchange for permission to publish the method.
3. Second-session testing as a product. Almost all internal training evaluations test once, on the day. Offer delayed retesting as an add-on service. Buyer: anyone running staff training with a budget to defend. First test: ask three training managers when they last measured retention a month later.
How sure can you be?
Be sure of very little, and that is the point.
In its favour: it is peer reviewed, published in a specialist surgical journal, it tested twice rather than once, and the authors have reported a result that does not flatter the technology they tested.
Against it: it is explicitly an exploratory pilot. The abstract does not state the number of participants, and a pilot study is usually very small. A small study can easily miss a real benefit — that is called being underpowered, and it is the most likely innocent explanation for these numbers. The abstract also does not say what the scores are out of, so the gap between 28.3 and 32.0 cannot be judged in context.
So the correct conclusion is narrow: this study did not detect a benefit. Not that there is none.
Watch for a properly powered trial with a stated participant count and a pre-registered plan. If one appears and also finds nothing, that is when the evidence has genuinely turned.
Do this today
Think of a technology you believe works. Try to remember a single study you have read that found it did not. If you cannot, you have only seen the filtered half.
Source: Mixed-reality-assisted training for vascular anastomosis: An exploratory pilot study, Annals of Vascular Surgery, October 2026. 10.1016/j.avsg.2026.10.037 · doi.org (peer reviewed).
Just Out Tech explains new research in plain language. This article was drafted with AI assistance and checked by a human against the original source.
- An exploratory pilot study published in Annals of Vascular Surgery (DOI 10.1016/j.avsg.2026.10.037) found no significant difference between groups trained with and without mixed reality assistance.
- First-session total scores were 28.3 plus or minus 3.1 and 32.0 plus or minus 2.5, with a p value of 0.106.
- Second-session total scores were 31.8 plus or minus 2.8 and 30.7 plus or minus 1.7, with a p value of 0.467.
- The groups did not separate further by the second session, which is what you would expect if the headset training had taught something durable.
- Because it is an exploratory pilot and the abstract gives no participant count, the most likely innocent explanation is that the study was too small to detect a real benefit.
The authors describe it as an exploratory pilot, which is a first small look run to decide whether a larger study is worthwhile.
Questions people ask
Does this prove mixed reality does not help surgical training?
No. One small pilot cannot show that, and the paper does not claim it.
What is publication bias?
Studies that find nothing are less likely to be written up and read, so the evidence you see about any technology is weighted towards the times it worked.
What does underpowered mean?
The study was probably too small to detect a real benefit even if one exists, which is the most likely harmless reason for these results.