Machine Learning Algorithm Guides Catalyst Choices for Magnesium-Catalyzed Asymmetric Reactions

Area of Science:

• Organic Chemistry
• Catalysis
• Machine Learning

Background:

• Vast organic-chemical literature details numerous catalysts and reactions.
• Current literature often documents catalyst scope without guaranteeing optimal performance (yield, enantiomeric excess).

Purpose of the Study:

• To develop a Machine Learning model for enhanced catalyst-reaction assignments.
• To improve the prediction of optimal catalysts for specific organic transformations.

Main Methods:

• Curated literature data was used to train a Machine Learning model.
• The model was applied to asymmetric magnesium catalysis.

Main Results:

• The Machine Learning model demonstrated relatively high accuracy in predicting catalyst-reaction pairings.
• Out-of-the-box predictions were successfully validated through experimental testing.

Conclusions:

• The developed Machine Learning model effectively improves catalyst-reaction assignments.
• The model shows promise for guiding experimental design in complex asymmetric synthesis, including reductions and Michael additions.