Integration of machine learning and digital twin in additive manufacturing of polymeric-based materials and products

Additive manufacturing (AM) has become a key enabler across industries, offering flexibility to produce complex, lightweight, and customized components. In recent years, machine learning (ML) has increasingly been adopted in AM to support tasks, such as predicting material behavior, detecting defects, and designing composites for specific performance targets. In parallel, digital twin (DiTw) technologies are gaining momentum as dynamic, real-time frameworks for process simulation, optimization, and predictive control. Polymeric materials and their composites are widely used in AM due to their strength-to-weight advantages, functional tunability, and ease of processing. One of the key reasons for the integration of ML in this domain is the anisotropy experienced in polymer AM, where mechanical and thermal properties vary with build direction, making this system an ideal candidate for data-driven modeling and optimization of adaptive processes. This review paper amalgamates the state-of-the-art developments at the intersection of ML, DiTw, and polymer-based AM. We investigated and compared the utilization of these technologies in the areas of manufacturing, parameter tuning, and product performance enhancement. The paper further outlines the key limitations and potential new applications, with some insight into how these might be considered in future research directions. In general, this work is intended to serve as a practical and future-oriented guide for researchers and practitioners working toward intelligent, data-augmented AM systems.