This study aims to critically evaluate recent advancements in intelligent textile sensors integrated with machine learning (ML) for continuous, non-invasive monitoring of athlete physiology. The focus is on enabling real-time, in-field assessment of biomechanical, biopotential and biochemical signals to optimize athletic performance and prevent injury.
This review systematically explores the material and engineering foundations of e-textile sensors, including conductive materials, sensing modalities and fabrication techniques. This study then examines the integration of ML pipelines for signal processing, feature extraction and model inference, with attention to deployment challenges such as data scarcity, algorithm generalizability and edge computing limitations. Real-world use cases across various sports contexts are analyzed to assess efficacy and identify translational bottlenecks.
Advanced conductive materials (e.g. AgNW composites, graphene and PEDOT:PSS hybrids) and fabrication methods (e.g. embroidery and inkjet printing) have enabled increasingly robust sensor integration into garments. Coupled with ML models – ranging from SVMs to CNN-LSTM hybrids – these systems can classify movement, detect fatigue and monitor cardiovascular and metabolic states. However, persistent gaps in real-world durability, signal noise under motion and ethical issues related to privacy and data ownership remain critical challenges.
This review uniquely bridges material science, wearable electronics, data science and ethics to present an interdisciplinary roadmap for next-generation athlete monitoring. This study introduces the concept of “Human Digital Twins” and outlines the convergence of multimodal sensor fusion, edge computing and self-powered systems as the future direction of smart sportswear. The findings of this study emphasize that technical breakthroughs must be matched by robust validation and governance frameworks to achieve meaningful adoption in elite and amateur sports alike.
