Wearable Sensor-Based Knee Joint Angle Estimation: Modalities, Modeling, and Applications

Wearable sensor-based Knee Joint Angle Estimation (KAE) supports rehabilitation assessment, assistive control, sports biomechanics, and remote mobility monitoring, yet current evidence is fragmented by heterogeneous sensor setups, modality combinations, data granularity, and validation protocols. This review synthesizes 128 peer-reviewed studies (2008-2024) and provides a KAE-specific perspective on application domains, sensing modalities (e.g., sEMG, IMUs, pressure, optical, and emerging flexible sensors), data sources (experimental/public/synthetic), and modeling approaches from classical regression to deep networks. We consolidate commonly used evaluation metrics (MAE, RMSE, R², NRMSE, PCC/CCC) and discuss their roles in routine monitoring versus safety-critical control. To improve reproducibility, we summarize key challenges with paper-level quantitative evidence (n/N, %) and offer actionable recommendations. Major barriers include reliance on controlled self-collected datasets, limited subject-independent testing, and insufficient robustness reporting under drift and placement variability. We outline directions toward standardized multimodal benchmarks, consistent validation, and clinically relevant testing for reliable translation.