New blog post on the DDMoRe Model Definition Language community website www.mdl.community. “…But I only use NONMEM”.
One of the most common responses I hear when I talk about the DDMoRe model exchange standards is “Why should I learn and use MDL? I only use NONMEM…” This may be true for you and according to the ISOP survey on model sharing it’s clearly true that the majority of respondents (mostly pharmacometricians?) use NONMEM. But think about the impact that the model you’re developing might have if it could be shared outside the pharmacometrics discipline - perhaps a statistician might want to use it to make some predictions or simulations of future trials to assess analytical methods or decision criteria for the trial. Are we going to insist they use NONMEM too?
My first job in Pharmacometrics was to take a NONMEM model and recode it in Pharsight’s Trial Designer (before the advent of Trial Simulator) in order to do just this kind of simulation and prediction for new dosage forms and PK profiles. It took quite a lot of work to ensure that the original NONMEM model corresponded to that in the simulator before I could move on and simulate outcomes for new PK profiles. That was ~15 years ago… Yet today, I’d probably be doing the same thing, but perhaps trying to use mrgsolve or an equivalent tool.
So, even though you might only use NONMEM, wouldn’t it be nice for your models to have wider impact? Be easier to share across quantitative disciplines? I hope that the advent of DDMoRe’s MDL and associated model exchange standards will help solve these problems.
…and hopefully within the next 15 years!!