Nonlinear mixed-effects models (may) consider data below the limit of quantification (BLQ) in parameter estimation. However, an evaluation of the goodness-of-fit plots (observations vs predictions in particular, using spline interpolation), displays a strong trend (of spline interpolation, but not of the data) in the region of censored data, as if the model disregarded BLQ data and the data were the lower limit of quantification itself, as structured in the database. I believe that the database is structured correctly and that the model considered the censored interval. Apparently this plot is the only one to exhibit this behavior.
Is spline interpolation adequately representing the competence/capability of the final model in this case? How to handle this situation?
Hi
It's a very good question. The use of a BLQ calibration curve is not only a question of the fit in this specific part of the calibration curve. That is why it is not enough to prove a linear fit in the LLOQ-HLOQ or LOD-HLOQ range to use an analytical method in practice. I think that what you are talking about is a very good direction. However, it seems to me that for this to make practical sense, you should try to adapt the methods you write about and try to propose a competitive model for the entire validation procedure. I think it would be a good idea to try to apply population methods to assess (validate) bioanalytical methods. However, it cannot be done randomly (only linearity evaluation in the LLOQ-HLOQ or LOD-HLOQ range, etc). This is because validation is (or should be) a holistic process. The use of population methods with properly defined covariates (and maybe DOE implementation) seems to be a much better approach for method validation than the current one. The current standards are based on the FDA and EMA guidelines, FDA published the first version of this guidance..... 20 years ago .... I'm sure its time to change philosophy...... :)
FDA
https://www.fda.gov/files/drugs/published/Bioanalytical-Method-Validation-Guidance-for-Industry.pdf
EMA
https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-bioanalytical-method-validation_en.pdf
Best regards
Tomasz
See these links:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6182903/
https://www.polyu.edu.hk/ama/staff/zhaoxq/Papers/Jiang-Ye-Zhao-Sinica2018.pdf
https://www.tesisenred.net/bitstream/handle/10803/6521/01Mcr01de01.pdf?sequence=1&isAllowed=y
Article A review of spline function procedures in R
Thank you Remal Al-Gounmeein . I will definitely check them out!
- Badges
- Science topic
- Similar topics
- Mathematics