Artificial intelligence for performance prediction of organic solvent nanofiltration membranes

There is an urgent need to develop predictive methodologies that will fast-track the industrial implementation of organic solvent nanofiltration (OSN). However, performance prediction OSN membranes has been a daunting and challenging task, due high number possible solvents complex relationship between solvent-membrane, solute-solvent, solute-membrane interactions. Therefore, instead developing fundamental mathematical equations, we have broken away from conventions by compiling large dataset building artificial intelligence (AI) based models for both rejection permeance, on collected containing 38,430 datapoints with more than 18 dimensions (parameters). To elucidate important parameters affect membrane performance, carried out thorough principal component analysis (PCA), which revealed factors affecting permeance are surprisingly similar. We trained three different AI (artificial neural network, support vector machine, random forest) predicted unprecedented accuracy, as 98% (permeance) 91% (rejection). Our findings pave way towards appropriate data standardization, not only prediction, but also better design development.