Avid Bioservices Inc (CDMO)

[Protector]

MazelMan, if you really KNOW it is linear you didn't understand the explanation because it is not linear.

BioBS has explained it in his post on the topic. If the centre enrolment is hockey stick and it was and we can PROVE it with clinicaltrials.gov archives, then centre enrolment was non-linear. As a consequence ANY next step based on that non-linearity is BY DEFINITION not linear.

Even if the enrolment assumption PER TREATMENT CENTRE is linear in the simulation (which has NEVER been under discussion and always in detail explained in every simulation) then the simulation REMAINS non-linear.

It is the branch jbainseky grabed to argue himself out of this because PPHM has no practical means to have a non linear patients enrolment PER TREATMENT CENTRE (double blinded - they are NOT involved). Only jbainseky's arguments that PPHM MIGHT, without telling us what factor in practice could drive that non-linearity, is insufficient to just say "one in x months, then one in y months, etc. It is a theoretical statement with no underlying practicable model UNLESS PPHM would have DICTATED the minimum/maximum #patients per time window (where a centre could NEVER guarantee a minimum and would have to let candidate go if reaching the maximum which would be square stupid).

On the other hand golfho and I had one non linear approach for the patients enrolment PER TREATMENT CETRE. Golfho used one simulation with an initial/upfront patient ready at opening of the centre and then one patient per time window (1 patient every x days). I choose NOT to use that although the possibility is in my simulator. See my very first simulation reports, the UPFRONT patient is in there.

I chose not to use it any more because posters on here that had some field collected information demonstrated that it was wrong to think the centres had a patient on opening and their field data confirmed it. So it was an educated choice and I let the computer recalculate golfho's simulation that did include such upfront patient into the straight # days equivalent (resulting in the famous 112 days consensus).

So if you or jbainseky can come up with ANOTHER non-linear model that is practical (meaning the model explains why there would FOR A SAME CENTRE be different spreads/waves of patient enrolment) that WOULD NOT be accommodated for in a simulation by using a linear equivalent based on patients per centre per day, then please be my guest. The more that my simulation reports always have the reminder that it are simulations, hence based on a number of assumptions and no EXACT prediction.

As for:

I know it is linear and most understand the meaning.

I think by know you must realize you didn't know and were merly under the impression you knew, while indeed most others understood the meaning and even posted about it to support the non-linear of the total simulation even when a linear enrolment scheme per treatment centre is used. Anyway, if after the above detailed explanation you still think it is non-linear then I rest my case for reasons of futility.

Not easy for some to admit being wrong but heck. It's a message board.

Apparently true, maybe at this point you and jbainseky may take under consideration that you might have been wrong and maybe admit it. That is what I have done several times in the past, and some other posters on occasion, so it is not impossible and there is no shame in it. We all make a best attempt to contribute, at least I think. Not that I expect or need any wrong admitting.