机器学习与无标记步态分析预测下肢损伤后重返运动的准确性：系统评价与Meta分析
简介
这项系统评价与Meta分析纳入了11项研究，评估了机器学习和无标记步态分析对下肢损伤（包括ACL损伤）后重返运动及再伤风险的预测准确性。结果显示，基于可穿戴传感器的机器学习模型预测重返运动的合并准确率为0.92，预测再伤风险的合并准确率为0.90，深度学习和集成模型优于传统方法。尽管准确性高，但方法学异质性和缺乏外部验证限制了其临床推广，亟需标准化报告和严格…
英文摘要
BACKGROUND AND OBJECTIVE: Anterior cruciate ligament and lower extremity injuries impose a substantial burden in sports medicine, yet conventional assessments fail to capture the dynamic nature of injury risk and recovery. Machine learning and markerless gait analysis offer potential improvements in predicting return-to-sport readiness and re-injury risk, but their clinical reliability remains uncertain. This review evaluates the predictive accuracy and clinical applicability of these approaches. METHODS: A comprehensive search of PubMed, Embase, Scopus, IEEE Xplore, and CINAHL was conducted up to March 2026. Studies involving adults with lower extremity injuries using machine learning or markerless systems and reporting predictive metrics were included. Pooled accuracy estimates were calculated using a fixed-effect model with logit transformation. Heterogeneity was assessed using the I² statistic. RESULTS: Eleven studies met inclusion criteria for quantitative synthesis. For return-to-sport readiness (six wearable sensor studies), pooled accuracy was 0.92 (95% CI: 0.89-0.95; I² = 52.2%). Sensitivity ranged from 0.79 to 0.99, and specificity from 0.75 to 0.99. For re-injury risk (two wearable sensor studies), pooled accuracy was 0.90 (95% CI: 0.84-0.95; I² = 0.0%). Markerless motion analysis demonstrated sensitivity of 0.82 and specificity of 0.77 for injury risk screening. Deep learning and ensemble models outperformed traditional approaches. CONCLUSION: Wearable sensor-based machine learning models achieve high predictive accuracy for return-to-sport and re-injury risk. However, methodological heterogeneity, reliance on internal validation, and absence of external validation limit generalisability. Standardised reporting and rigorous external validation are urgently needed before clinical implementation. SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261352158, identifier CRD420261352158.