Interpretable Prediction of Aggregation-Induced Emission Molecules Based on Graph Neural Network

We developed an interpretable graph neural network (96.4% accuracy) for AIEgens identification, revealing 24 characteristic functional groups. Based on these insights, two virtual library strategies (self-fragment and donor-acceptor docking) were proposed and predicted four experimentally confirmed AIEgens successfully, which establishes a rational design framework for AIE materials.

Read more...