Discussion about this post

User's avatar
Erik's avatar

I used to work with system identification quite a bit in the past, using input design even, it was great fun and worked pretty well. I think many would hesitate to have the same enthusiasm for medicine, where even the "simple binary" often is too difficult, and the systems are evolving all the time, and placebo/nocebo exist etc. As you point out, the trick here is to identify candidate responses where this particular setup could be considered (timescales, no lingering system memory, ...) but also where it may matter (clinical relevance) since some things being measured can be oversold as important. Single variable responses that are clear are perhaps few and far between in the human body which itself is running in full chemical feedback mode all the time (homeostasis). Anyway, it is great fun to follow your series of posts here!

Expand full comment
1 more comment...

No posts