I have this PK/PD model (user defined) with very rich PK PD data. Initially it all worked fine. When I tried to use a more complicated model (e.g. adding a precursor compartment), then I got “subscript out of range” error for one individual. But the model still finished the run and returned results. The individual in question does not have prediction but still with parameters estimated. What could be the problem? Can I still trust the results? Thanks. Jun
Dear Jun, Are you using WinNonlin 5.x or earlier, or are you using Phoenix? Emily
Hi, Emily, I am on WinNonlin 5.2.1 Jun
Can you post the code?
COMMANDS NFUNCTIONS 2 NDERIVATIVES 5 NPARAMETERS 15 PNAMES ‘CL’, ‘Vc’, ‘K12’, ‘K21’, ‘K13’, ‘K31’, ‘Kout’, ‘R0’, ‘Emax’, ‘EC50’, ‘AMP’, ‘PERIOD’, ‘ACR’, ‘Rp’,‘Gamma’ PUNIT ‘L/hr’,‘L’,‘1/hr’,‘1/hr’,‘1/hr’,‘1/hr’,‘1/hr’,‘pg/mL’,‘~’,‘ug/L’,‘pg/mL’,‘hr’,‘hr’,‘pg/mL’,‘~’ NSECONDARY 2 SNAMES ‘Kp_in’,‘Kp_out’ SUNIT ‘pg/mL/hr’,‘1/hr’ END remark - define temporary variables TEMPORARY t=x Dose=CON(1) Kp_in=KoutR0 Kp_out=Kp_in/Rp Drug=1-Emax(Z(1)/Vc)Gamma/(EC50Gamma+(Z(1)/Vc)*Gamma) END remark - define differential equations starting values START Z(1) = Dose Z(2) = 0 Z(3) = 0 Z(4) = Rp Z(5) = R0 END remark - define differential equations DIFFERENTIAL DZ(1) = -(K12+K13+CL/Vc)Z(1)+K21Z(2)+K31Z(3) DZ(2) = K12Z(1)-K21Z(2) DZ(3) = K13Z(1)-K31Z(3) DZ(4) = Kp_in - Kp_outDrugZ(4) DZ(5) = Kp_outDrugZ(4)-KoutZ(5) END remark - define algebraic functions FUNCTION 1 F= Z(1)/Vc END FUNCTION 2 F= Z(5)+ AMPsin(23.1415((T+ACR)/PERIOD)) END remark - define any secondary parameters SECO Kp_in=Kout*R0 Kp_out=Kp_in/Rp END remark - end of model EOM
Are there any blank cells in the dosing section? For example, if you entered there would be two doses, but only entered information for one dose, or left a cell blank? Is there anything different about the first subject’s data compared with the rest of the subjects? “Subscript out of range” usually means there was a matrix defined somewhere in the program, and the input given goes outside the dimensions of that matrix. This could be an internal matrix (such as built in functions, in which case the problem can be resolved by fixing the model code that uses those functions) or one that you have access to through the UI (e.g., data worksheets and other user inputs given in the UI).
Another thing you might try is taking out the underscores (e.g., kpin instead of kp_in).
Emily, Thanks for the suggestions. There is no empty cells in dosing regimen. Actually I used the exact same dataset for other simpler models just fine (same dosing regimen). I tried to delete the underscore it didn’t work either. And every time it always had the problem with same individual and I checked there was nothing special about that guy. Thanks. Jun