model/nonmem/1009.mod

[jenna-a2ai]

NONMEM Model

Note: Please modify the checklist items to insert relevant QC context.

NONMEM Model Control Stream

• 🔴 Correct data files are used in model.
  • correct data, file name and hash: [INSERT]
• $INPUT matches Data columns.
  • If columns are ignored, explain which ones (i.e. character columns at the end of a dataset): [INSERT]
• IGNORE statements subset the data correctly.
  • What are the IGNORE statements that SHOULD be there and why: [INSERT]
• Correct TIME and DV variables.
  • DV or transformed DV used: [INSERT]
  • time column: [INSERT]
• Number of THETAs in the model match in $THETA.
• Covariates on appropriate THETA.
  • covariates and the parameters they SHOULD be on and functional form: [INSERT]
• Correct Scale Factor and conversions used.
  • what the equation should be and why (e.g. S2 = V2/1000 because dose in mg and conc in ng/mL): [INSERT]
• ETAs on correct parameters and match labels in $OMEGA block.
• Residual Error model coded correctly.
  • confirm error model (i.e., proportional error): [INSERT]
• Description of PK model matches ADVAN/$DES.
  • PK model description (i.e., 2-CMT model with first-order absorption and LAG time): [INSERT]
• If Simulation, SEED included in $SIM.
• $EST method consistent with expectation.
  • expectation (I.e., FOCE-I): [INSERT]
• $COV step included for model runs, include MATRIX.

NONMEM Model Output

• Check Run Number matches the control stream.
• Successful $EST step.
  • MINIMIZATION SUCCESSFUL.
  • Absence of warnings like MINIMIZATION TERMINATED or ROUNDING ERRORS.
  • Check for any warnings or errors, such as SIGDIGIT, NO. OF FUNCTION EVALUATIONS EXCEEDED, or any numerical issues that might affect the model.
• Successful $COV step.
  • Ensure the matrix is positive definite. A non-positive definite matrix may indicate issues with model estimation.
• Check parameters are not highly correlated.
  • In the CORRELATION MATRIX OF ESTIMATE section, correlation coefficients <0.95.
• Condition Number < 1000.
  • Found in the EIGENVALUES OF THE COVARIANCE MATRIX section, largest eigenvalue divided by smallest eigenvalue.
• List parameters with high shrinkage (>30%) to flag for analyst/medical writer.

Metadata

• author: jenna-a2ai [email protected]
• qc type: NONMEM Model
• script hash: 88909315f4b0b2265faa6afe11c2fdc5
• git sha: 3bacb81
• file history: https://github.com/A2-ai/gilead_ghqc_demo/commits/main/model/nonmem/1009.mod