A software‑supported, AI‑assisted workflow for optimising LC gradient methods

The tight timelines that all pharmaceutical laboratories face must consistently be overcome if they are to deliver validation-ready liquid chromatography (LC) methods.

Yet, method development often proves to be particularly time consuming. Its reliance on manual trial-and-error, alongside the need to create multiple analysis methods, then run and then interpret the ensuing chromatograms and then decide on the next adjustment all combine to create a labour-intensive process.

Although this kind of iterative approach can certainly be effective, it can also introduce variability – and its success can depend heavily on an individual analyst’s experience and judgement.

In a new article, Automated AI-driven HPLC method development, a team of experts from Shimadzu introduce a novel automated HPLC method development that is driven by artificial intelligence (AI) and could speed up the process and improve its accuracy.

The algorithm that the authors describe can be implemented with dedicated method-development software and used to automate the exploration and refinement of gradient conditions.

Presenting a case study on the process, the article shows that a neutral, criteria‑driven workflow can converge rapidly on robust separation conditions while also reducing the need for empirical trial‑and‑error. Consequently, use of the algorithm can support the development, transfer and lifecycle management of LC methods.

Beyond the seven-component mixture featured in the case study, the new workflow is also shown to have a generalised application to pharmaceutical tasks where robust separation and efficient runtime are required.

The authors conclude: “By articulating clear criteria and allowing the algorithm to manipulate gradient shape, method developers can reduce manual iterations, document a transparent path to the final conditions, and improve the consistency of outcomes across analysts and sites.”

Read the full article here