Why there are always 5 points with the prediciton much less than the actual concentraion

Phoenix Certara Forums Pirana and RsNLME
1

Hi,

I am new to NONMEM and I am trying to build a model. I attempted both the one-compartment model and two-compartment model with different residual error models (additive, proportional, combined and exponential). The diagnostic graph all look nice with the gradient and R2 close to 1. Still, there are always 5 or 6 stubborn points with the populaiton prediction (PRED) much less than the actual concentration. I checked the time-concentraion curce and it turned out that these points are normal. I have no idea what to do. Anyone who can give me a hand would be my saver! Thank youuuu guys!

Best regards

1.png
1.png480×288 3.99 KB

2

Hi Fafa, strictly speaking this is a community for NLME, Pirana etc,. (i.e.Certara products), however to look at your modelling questuion we woudl really need to see raw data and NM control stream and any other model you have created.

How do your residual and overall GoF plots look ?

3

Ohhh! Guess I come to the wrong place. I feel sooo sorry Simon :frowning: But still, thank you for your kind help!!

I think I’ll go somewhere else to solve it.

P.S I was wondering if you could share a plateform to discuss the problems in NONMEM :slight_smile: (Just ask hh

Best regards!

4

Join NM user list that is hosted here; https://www.cognigen.com/nmusers/thread.html

but you can’t add attachments there so happy to take a look here too.

You never know you may find yourslef a convert to the ease of https://www.certara.com/software/r-speaks-nlme-rsnlme/ !

5

Thank youuuuuu Simon for your kind help! Best wishes~

I will talk to my boss about Certara :slight_smile: