ML laboratory: machine learning software for formulation optimisation and experimental planning

ML Laboratory is software developed by SIA Solven to help R&D teams use experimental data to predict properties, explore formulations and plan further experiments. It is designed for chemistry, materials development and process optimisation, where teams need to understand how composition and processing conditions affect measured results.

The workflow combines data quality checks, predictive modelling, model evaluation, experimental planning, optimisation and report generation. Researchers can use existing datasets to investigate relationships between inputs and outcomes, compare candidate formulations or process settings, and identify promising conditions for experimental testing. Predictions support decisions and require confirmation through experiments.

An internal end-to-end test on a synthetic coating-adhesion dataset completed the workflow from project setup to report generation, including predictions, planning and optimisation outputs. This supports our current TRL 4 assessment: validation in controlled conditions. The next step is evaluation on a partner’s real experimental data and R&D workflow.

We are seeking R&D teams interested in software licensing, pilot or trial engagements, and co-development for specific applications. A pilot would focus on one defined problem, an agreed dataset and measurable evaluation criteria, such as predictive accuracy and the usefulness of suggested experiments.

Potential applications include formulation development, prediction of material properties and optimisation of processing conditions.