Jump to content


Photo

Closed form vs differential equation


  • Please log in to reply
2 replies to this topic

#1 cmukashyaka

cmukashyaka

    Newbie

  • Members
  • Pip
  • 3 posts

Posted 30 November 2023 - 04:30 PM

*
POPULAR

Hello,

I am running a 2-compartment linear model, when I don't click closed form to use differential equations, the model is not converging with error message:

 

Initial parameter values result in -LL = NaN

 Restart from new values
Error: Model Exception: Fortran Exception
NlmeErrorExit: close files and exit(0).
 
Additionally, initial estimates graphs have some subjects with negative concentrations. I change initial estimates to no avail. 
 
However, when I run the exact same model in closed form, the model runs fine and I get descent fits and parameter estimates with good CV%. 
 
Anyone else experienced this? 

Attached Files


  • WalterNic, ??????????????? ??????, DanielGon and 67 others like this

#2 kniefort

kniefort

    Member

  • Val_Members
  • PipPip
  • 29 posts

Posted 30 November 2023 - 05:44 PM

I note in your dataset you have many rows with Conc=NA, some with duplicate time data for existing observations.  It's always good practice to prepare "clean" datasets, removing rows that do not contain any useful information for your analysis objectives.  If you filter your dataset to remove the NA/duplicate concentrations, you will find that you get the same results for closed vs open format.


  • Stevensi likes this

#3 cmukashyaka

cmukashyaka

    Newbie

  • Members
  • Pip
  • 3 posts

Posted 12 December 2023 - 08:50 PM

 

I note in your dataset you have many rows with Conc=NA, some with duplicate time data for existing observations.  It's always good practice to prepare "clean" datasets, removing rows that do not contain any useful information for your analysis objectives.  If you filter your dataset to remove the NA/duplicate concentrations, you will find that you get the same results for closed vs open format.

 

Thank you. I will clean up the data and see if that helps






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users