Using machine learning, MIT chemical engineers created a computational model that can predict how well a given molecule will dissolve in an organic solvent. This type of prediction could make it much easier to develop new ways to produce pharmaceuticals and other useful molecules.