
Simple model generation with differentiation between similar reaction schemes and extraction of core kinetic parameters.
Initial experiment suggestions, followed by targeted differentiation.

Evaluate models and extract core parameters such as activation energy.



No, the algorithm automatically proposes reaction schemes, and ranks them based on complexity and goodness of fit. Custom models can be proposed and compared to algorithmically generated models.
Goodness of fit (R2) and model complexity (MAE) are combined into one metric which is used to rank the models. This ensures that the model is well-fitted, while preventing overfitting. Where two models are very similar, differentiation experiments can be proposed to identify the correct model.
Simple time-series data is used. Temporal measurements of concentration, with runs at multiple temperatures and different starting concentrations is enough.
Yes, we support direct integration with hardware for automated data generation. We natively integrate with ReactAll. Speak to our team for more information, or custom integrations.