JAREX: An Acquisition Function for Multi-Objective Algorithmic Process Characterization
JAREX maps multi-objective pharmaceutical process pass regions with fewer experiments than factorial design.
The paper presents JAREX, a Bayesian active-learning acquisition function for multi-objective pharmaceutical process characterization under Quality by Design. It selects experiments to recover the joint pass region where several threshold criteria are satisfied at once, combining an optimistic feasibility mask with a multi-objective randomized straddle. Benchmarks report more accurate, sample-efficient recovery than factorial design of experiments, space-filling designs, and greedy strategies, and batched use reduced iterative experiments by more than half. The method is implemented in the open-source obsidian package.
- Learns the joint pass region across multiple quality thresholds
- Outperformed factorial DOE, space-filling, and greedy objective-wise designs
- Batched use cut iterative characterization experiments by more than half
- Implemented in the open-source obsidian package
- Tags
- bayesian-optimizationactive-learningprocess-characterizationpharmaceuticalsquality-by-designacquisition-function
- Products
- obsidian
From the original article
1 min read at arxiv.orgPharmaceutical process characterization is central to Quality by Design because it defines how variations in process parameters affect the ability to meet product quality specifications, thereby supporting proven acceptable ranges and robust manufacturing.
By Xinyang Li · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.