I think you missed my point. (But I checked your link). It...

I think you missed my point. (But I checked your link). It matters that the change occurred a long way into the trial (when some patients had already been treated in both groups thus (group a - pre-change, group b pre-change) - because it essentially introduces a new split - those treated before and after the protocol change (group a - post change and group b - post change) - those before in the control arm may have been getting their wounds changed daily by a nurse (probably not the treatment arm (as keeping the cells close for a time is the point and the nurse was probably made clear about that)- so arguably getting better treatment (ie group b - no cells - daily change compared to group b - multiple daily change) and that better treatment (group b before versus group b after) is another variable over not getting it.

Speaking generally -

If you have two classes of outcomes - outcomes from set A (where say 85% reduction in wound size occurs on average) and outcomes from set B (where say 54% of reduction in wound size occurs on average) you still don't have enough information to be sure set A is better than set B statistically because if the treatment - it could be that set A (consider the extreme example of if there is only one in each set) just happens to have a patient that is very resilient irrespective of the treatment - they are a naturally better healer).

Any one can put up an observation that 85% reduction is more than 54% reduction. But its another thing entirely to say the reduction is STATISTICALLY SIGNIFICANT - those extra words require extra handling - and they STATISTICALLY SIGNIFICANT were not used in the reports on DFU to date.

You also need the know the sample sizes of the two sets. Otherwise you might have had just one value in each of the sets. Or a couple - now we know in this case it should be about 8 by the end.

Generally, in stats you need a sample size of about 6 as a minimum - from memory T tests etc. More - bigger sample sizes is better for confidence you've got enough mathematical power to reduce the chance variation. Because although a statistical comparison between A and B doesn't need to posit mechanisms of action its just trying to determine the probabilities or likelihoods of the variations in outcomes occurring by chance bigger groups are better.

When you have small sample sizes any change in the protocol part way through cuts in - because its another variable.

------------

Kilian can throw stuff out to the general population (who aren't going to be making deals to take on an indication anyway) without mentioning statistical significance - and they'll lap it up - but those who won't lap it up - because they understand stats are the ones that need to be impressed to make a deal in the first place.

----

Another point - because I'm replying to pfeifer - in the announcement with preso Kilian included a slide with mechanisms of action - and a reference to the paper he'd got the figure from.

He'd altered the figure though. The original paper was comparing different types of mscs - adipose, bone marrow etc. And it was bone marrow mscs not mscs that were in the original figure.

It certainly wasn't comparing iPSC derived MSCs. Kilian fudges over that. Knuckle head ignoramuses won't care - but the aren't the potential partners - potential partners won't fall for that sort of superficial fluff.

---

What is a cell - its a moving inside the body thing with a layer that receives signals from the enviroment and produced product (cytokines) and puts those out into the enviroment. Yes in the body cells - cheomtaxic cells move through various environments homing in on chemical signals something like how ants follow a chemical trail laid down by other ants.

You could say a cell is alive - like a beetle - but if you stick a live beetle to say a playing card with super glue say (attach a cell to a bandage) you aren't going to get natural cell mobility anymore than you get natural beetle mobility.