Dear Members,
I’m working on a PopPK model in Phoenix NLME using a dataset of 86 subjects, including both SAD and MAD data. Here’s a summary of my workflow:
Verified all numeric and textual data.
Filtered the dataset using DVID, then copied and saved the filtered data for model fitting.
Fitted a model with nonlinear absorption using textual model modifications.
Model fitting went smoothly, results generated , and I saved the project.
However, after shutting down my computer and reopening the project the next day, I noticed that certain subject data rows had moved—the dataset appeared manipulated, and same model is not getting re-run.
For issue identification, i repeated model building. Model fitted , While Phoenix was still open, everything worked fine, including duplicating and modifying the model. Both the original and filtered datasets were intact.
But after restarting the system, the dataset showed unexpected row reordering. I cannot share the dataset due to confidentiality, but I’d like to understand:
Has anyone experienced similar issues with Phoenix NLME?
Could this be related to how Phoenix saves or caches data?
Is there a known bug or workaround to prevent this kind of data manipulation after reopening?
Any insights or suggestions would be greatly appreciated.
Thank you
Hi Bhim, that does sound odd - if you can try to make a project that demonstrates the issue that you feel you can share with Support then we can investigate further.
However I would like to pick up on one of your steps - “copied and saved the filtered data”, if possible i would skip this set, there is no need to make a copy to a data folder normally - it is better to keep it in the workflow so it is intrinsically refreshed.
Dear SimonDavis,
This is a dummy dataset, where the yellow-highlighted cells indicate the use of “Alt+Shift” to insert a line break within a cell. Initially, Phoenix correctly interpreted these entries during data upload and model fitting. However, after saving and reopening the project, Phoenix would create an additional row in the dataset, placing the cell content after the next row of the same cell.
This issue arose because, in Excel, some of these entries appeared as single-row data, but actually contained multiple lines that were only visible upon double-clicking the cell. The dataset has now been corrected and is functioning properly.
To prevent such issues in the future, an automatic detection system during the initial dataset upload from Excel could be implemented as a secondary safeguard. Nevertheless, ensuring the dataset is correctly prepared remains the primary corrective action.
Thank you.
