Machine Learning for Wearable Sensor-Based Human Movement Rehabilitation: A Five-Year Systematic Review

Document Type

Article

Source of Publication

IEEE Access

Publication Date

1-1-2026

Abstract

Wearable-sensor-based human movement analysis is an increasingly important component of digital health and rehabilitation, enabling objective monitoring and data-driven personalization of therapy. In parallel, machine learning (ML) methods have rapidly expanded for interpreting multimodal movement signals, yet the evidence base remains heterogeneous and difficult to benchmark. This PRISMA-guided systematic review synthesizes recent ML approaches for wearable human motion analysis in rehabilitation-oriented health applications. We searched IEEE Xplore, PubMed, and Scopus for English-language studies published from 2021 to 2025 and extracted information on sensor modalities, ML task formulations and model families, dataset characteristics, validation protocols, and reported performance metrics, together with methodological quality indicators. A total of 174 studies met the inclusion criteria, spanning gait and lower-limb analysis, upper-limb rehabilitation, balance/fall-risk assessment, activity monitoring, and motor recovery prediction. The literature is dominated by inertial sensing (Iμaccelerometer) and, to a lesser extent, EMG-based systems, with models primarily formulated as supervised classification problems. Frequently used methods include SVM and Random Forest alongside deep architectures such as CNNs and LSTMs (including hybrid variants). Across studies, evaluation commonly relies on within-dataset cross-validation and a narrow set of scalar metrics (e.g., accuracy and F1-score), while external validation and standardized benchmarking remain comparatively rare. Despite strong reported performance in controlled settings, we identify persistent informatics and translational limitations, including small and private datasets, heterogeneous protocols, risks of biased or non-subject-wise validation, limited transparency, and sparse evidence of real-world clinical integration. We conclude with actionable priorities for trustworthy wearable ML in health: improved validation rigor and reporting standards, broader use of public benchmarks, explainable and patient-centered modeling, and multimodal, adaptive systems to support deployment in real-world telerehabilitation.

ISSN

2169-3536

Publisher

Institute of Electrical and Electronics Engineers (IEEE)

Volume

14

First Page

99280

Last Page

99307

Disciplines

Computer Sciences | Medicine and Health Sciences

Keywords

Biomedical data analytics, deep learning, gait, human movement analysis, inertial measurement units (IMU), machine learning, motion sensors, rehabilitation, systematic review, wearable health monitoring

Scopus ID

105043509079

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Indexed in Scopus

yes

Open Access

yes

Open Access Type

Gold: This publication is openly available in an open access journal/series

This document is currently not available here.

Share

COinS