NONMEM Model

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

[ ] :red_circle: 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.

[ ] 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 jenna@a2-ai.com
qc type: NONMEM Model
script hash: 88909315f4b0b2265faa6afe11c2fdc5
git sha: aced09b5e323d40eeab922726cccb1d06ed00cd2
file history: https://github.com/A2-ai/gilead_ghqc_demo/commits/main/model/nonmem/1009.mod