Polymer informatics for decision-making in industrial R&D: Challenges, opportunities, and perspectives

Polymer informatics has accelerated property prediction and materials exploration. However, in industrial R&D, predictive accuracy alone does not necessarily drive experimental decisions. We suggest that the key challenge lies in bridging predictions and decisions on the next step in the experimental workflow under limited resources. This article considers polymer informatics not only as a predictive technology but also as a decision-support technology for experimental action. We organize prior studies from three perspectives, data, understanding, and action, that support translating predictions into experimental action. The data perspective addresses data accumulation and utilization under the small-data and nonstationary conditions typical of functional materials development. The understanding perspective examines material-centric interpretability, grounded in physical properties and chemical structure, as a basis for rational design decisions. The action perspective discusses whether proposed candidates are experimentally executable, accounting for synthetic feasibility, raw material availability, and safety constraints. Through these perspectives, we outline pathways toward deploying polymer informatics as a practical technology that bridges prediction and experimental action in industrial R&D. Conceptual overview of the data, understanding, and action perspectives for polymer informatics driving experimental decision-making in industrial R&D.