The Polymer Society of Korea

Abstract

Biomedical polymers are required to simultaneously satisfy diverse performance criteria, including biocompatibility, mechanical properties, degradation behavior, and surface characteristics. These properties are intricately governed by polymer composition, molecular structure, and processing conditions, rendering conventional trial-and-error optimization approaches inefficient and time-consuming. In this study, an artificial intelligence (AI)-based framework is proposed for effective property control of various biomedical polymers. Comprehensive datasets were constructed using polymer composition, molecular parameters, and processing variables as input features, while mechanical, thermal, degradation, and biological properties were employed as output targets. Machine learning and deep learning models were applied to predict polymer properties and to identify critical structure–property relationships. The AI models successfully captured complex nonlinear correlations across different polymer systems and demonstrated high agreement with experimental results. Furthermore, inverse design strategies enabled the determination of optimal polymer compositions and processing conditions that satisfy target performance requirements. This AI-driven approach offers an efficient pathway for biomedical polymer design and is expected to accelerate the development of advanced biomaterials and customized biomedical polymer systems.