AI-integrated wearable systems achieved 86-95% classification accuracy for sports injury prediction, while workload-based models showed a 15-fold increase in injury risk at high workload ratios.
Systematic Review (n=15)
Do wearable sensors combined with AI/ML models accurately predict sports injuries and monitor rehabilitation in physically active individuals?
AI-integrated wearable systems show high accuracy for sports injury prediction and monitoring, though current evidence is limited by small sample sizes and lack of external validation.
Background and Study Aim. Wearable sensor technologies integrated with artificial intelligence (AI) and machine learning (ML) are increasingly used for continuous athletic monitoring. Despite their application, continuous monitoring of physically active populations in relation to injury susceptibility remains insufficiently implemented. This review synthesizes evidence on AI-integrated wearable systems for sports injury prediction and rehabilitation monitoring, emphasizing diagnostic performance and applicability to university physical education settings. Materials and Methods. PubMed, Web of Science, Scopus, and ProQuest were systematically searched through December 2025 following PRISMA 2020 guidelines (PROSPERO: CRD42026130911). Studies involving athletes or physically active individuals using wearable sensors combined with AI/ML models (CNN, LSTM, RNN) for injury prediction or rehabilitation monitoring were eligible. Risk of bias was assessed using PROBAST. Results. From 1,004 identified records, 15 studies met the inclusion criteria. Six studies explicitly involved university-aged participants (18–29 years) or were conducted in university laboratory settings. IMUs were the predominant sensor modality. Deep learning models (CNN, LSTM) achieved classification accuracies of 86–95% and F1 scores exceeding 0.90. Workload-based models demonstrated a 15-fold increase in injury risk at acute-to-chronic workload ratios above 1.27. PROBAST assessment identified only two studies as having a low overall risk of bias. Four studies were at high risk, primarily due to small samples, absent external validation, and analytical limitations. Conclusions. AI-integrated wearable systems show considerable potential for injury monitoring in athletic and university physical education contexts. However, small samples, limited external validation, and heterogeneous injury definitions constrain the current evidence. Future research should prioritize multicenter prospective studies and explicitly target university student populations.
Sharma et al. (Mon,) conducted a systematic review in Sports injury (n=15). AI-integrated wearable sensors was evaluated on Diagnostic performance for injury prediction and rehabilitation monitoring. AI-integrated wearable systems achieved 86-95% classification accuracy for sports injury prediction, while workload-based models showed a 15-fold increase in injury risk at high workload ratios.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: