How do I create a model with a defined profile as a factor?

[#1]

Hi all,

I have a question regarding DoE design. 

Normally, I use continuous factors with fixed values for 0, -1, +1, with the custom design DoE. However, I now have to create a model for a cell culture process (duration is multiple days), where one of the input factors has a profile at the lower range (the other factors will be fixed values for 0, -1, +1). Can someone explain to me how I can design this model? 

further more, what would be preferred, from a statistical point of view: 

1) keep upper range fixed and centerpoint fixed, so only profile for lower range

or

2) have same shape profile for upper range and centerpoint as used for -1

I have plotted scenenario 1 and 2 in the attached file as I cannot add the pictures in this message. 

any help is very welcome, I very much look forward to your response/ideas

kind regards,

Marjolein

[#2]

Hello Marjolein,

as far as I know, there is no preferred profile from the "statistical point of view". You have a categoric factor which is the profile. The question is: Which version does make more sense for the cell culture process and the purpose of your experiment?

Regards,
Andrea

[#3]

Hi Andrea,

Thanks for your reply! 

Can you explain a bit further how you would apply categorical variables to make the profile in time? I assumed that the levels for the categorical factors will be same value throughout process time, so its then not creating a profile over time.

kind regards, Marjolein

[#4 · Accepted solution]

Hi Marjolein,

yes, the categorical factor wouldn't follow a profile, but it doesn't have to. Just create a categorical factor with three levels:

Factor Name	Role	Values
X	Categorical	L1 = profile 1	L2 = profile 2	L3 = profile 3

The profiles don't have to be the same, each can follow an unique time-program. The "mid-level" (profile 2) doesn't have to be the "center-point" as you showed in your Word-file in figure 2. You can take the profiles as in figure 1.

Regards,

Andrea