Researchers at ETH Zurich and Roche Pharma have developed an advanced closed-loop workflow that enhances chemical reaction prediction, particularly in data-scarce drug discovery environments. This innovation integrates active and geometric deep learning techniques to improve the accuracy and efficiency of predicting chemical reactions. Concurrently, multiple studies have demonstrated the application of machine learning and computational chemistry across diverse areas such as heterogeneous catalysis, perovskite materials, low-k dielectric materials, and metal-organic frameworks. Notable advances include the use of convolutional neural networks and transformer architectures for modeling nonradiative recombination in CsPbI3 and Ge-doped perovskites, machine learning potentials capturing diffusion and ring-opening reactions, and AI frameworks optimizing synthesis sequences and reaction conditions in autonomous laboratories. These developments collectively illustrate the growing role of machine learning in accelerating material discovery, drug-target interaction prediction, and catalytic process design.