The Role of Artificial Intelligence and Machine Learning in Balance Classification Using Center of Pressure – A Comprehensive Review

Balance control is essential for safe daily activities, and impaired postural stability is strongly associated with fall risk, particularly in older adults. This systematic review examined the performance of artificial intelligence (AI) and machine learning (ML) methods for balance classification using Center of Pressure (CoP) data. Following PRISMA guidelines, 47 studies published between 2015 and 2025 were included. Force-platform-based post-urography remained the predominant sensing modality, while wearable technologies increasingly enabled assessment beyond laboratory environments. Support vector machine (SVM) was the most frequently used algorithm, followed by neural network (NN)-based models. Deep learning (DL) architectures, including convolutional neural networks (CNN) and long shortterm memory networks (LSTM), achieved high classification accuracy by capturing complex balance patterns. Traditional ML models also demonstrated strong performance with lower computational demands and greater interpretability, supporting clinical feasibility. However, class imbalance remains a key limitation, often reducing sensitivity in high-risk groups. Future studies should prioritize robust and clinically deployable models for reliable realworld balance assessment and personalized rehabilitation.