Key points are not available for this paper at this time.
Abstract The field of polymer science experiences a significant transformation through data‐driven approaches, automated synthesis, and machine learning (ML) systems, which create next‐generation polymers that form the core of polymer informatics. In this study, we evaluated polymer informatics through its predictive modeling, inverse design, and synthesis techniques. The technology has been discussed in detail, including its applications in materials development, the sustainable circular economy, polymer creation, biomaterials production, and additive manufacturing optimization. The combination of multitask deep neural networks, graph neural networks, and transformer‐based architectures with SMILES strings and molecular graphs has enabled ML algorithms to predict thermal, mechanical, electrical, and optical properties with high performance. The field requires standardized data as its foundation, sourced from major databases (e.g., PoLyInfo and PI1M) to improve property predictions in computational systems. Researchers employ graph‐based models, topological indices, and ML to design polymers via optimized closed‐loop systems that track processes in real time. The development of polymer informatics faces two primary obstacles: researchers lack access to sufficient standardized, high‐quality datasets, and they struggle to represent polymer structures effectively (e.g., BigSMILES). The upcoming development of polymer informatics will focus on self‐driving laboratories, AI‐based retrosynthesis, and large language models for synthesis planning.
Islam et al. (Sun,) studied this question.