AI/ML Now: Uncertainty-Driven Model-Based Search Methods for Method Development in Liquid Chromatography (Part One)

In this episode of AI/ML Now, we talk with Leon Niezen and Gert Desmet of the Vrije Universiteit Brussel about how predictive uncertainty can make automated LC method development more reliable. In model-based method development, retention models are often refined by running each new experiment wherever the chromatographic response function (CRF) looks best. Niezen and Desmet took a different route: they gave the retention models a predictive uncertainty function and used the uncertainty itself to choose the next run.

They tested the approach in silico on 122 randomized mixtures, comparing a conventional CRF-driven search with two uncertainty-driven strategies, one global and one per analyte. Once peak overlap made peaks hard to detect, the uncertainty-driven methods did clearly better. When a resolution of 0.5 was needed to track peaks, they failed for 1 sample in 122, while the CRF-driven search failed in 14% of cases. The work was published in the Journal of Chromatography A in 2026 (1).

Desmet and Niezen answered the following questions:

• What was the behind your published paper” “ Uncertainty-Driven Model-Based Search Methods for Method Development in Liquid Chromatography”?1
• The uncertainty-driven strategies performed particularly well when peak detectability was hindered by peak overlap. Why does reducing model uncertainty become particularly important under these conditions?

Watch the full video here:

View Part Two here:

AI/ML Now: Uncertainty-Driven Model-Based Search Methods for Method Development in Liquid Chromatography (Part Two)

References

1. Niezen, L. E.; Cabooter, D.; Desmet, G. Uncertainty-Driven Model-Based Search Methods for Method Development in Liquid Chromatography. J. Chromatogr. A 2026, 1781, 467071. DOI: 10.1016/j.chroma.2026.467071.
2. Uncertainty-Driven Model-Based Search Methods for Method Development in Liquid Chromatography. Presentation at ISC 2026, Prague, Czech Republic.