长距离跑后“足外翻”姿态增加膝关节内侧接触力：基于OpenSim肌骨建模及机器学习预测的研究

Foot Pronation after Prolonged Running Increased the Medial Contact Force in the Knee Joint:A Study Based on OpenSim Modelling and Machine Learning Prediction

目的:揭示跑者在长距离跑后下肢关节的负荷与足姿态的变化及膝关节接触力分布且验证二者关联,为快速预测膝关节负荷提供简易手段。方法:运用OpenSim个体化肌骨建模分析业余跑者5 km跑后下肢关节的运动学、动力学及膝关节接触力变化,采用一维统计参数映射(Statistical Parametric Mapping 1d)分析法检测差异性,结合足姿态参数与膝关节总接触力及膝关节内外侧接触力训练并检验机器学习算法。结果:1)5 km跑后,髋关节蹬离期伸角度、支撑初期外旋角度及踝关节蹬离期跖屈角度均显著增大;2)髋关节支撑期伸力矩及支撑初期外展力矩、踝关节支撑期跖屈力矩均显著增大,膝关节支撑初期及蹬离期屈力矩均显著降低;3)膝关节蹬离期总接触力显著减小,支撑初期内侧接触力显著增大而支撑初期及蹬离期外侧接触力均显著减小;4)机器学习算法模型发现,足外翻程度增大引起膝关节总接触力及内侧接触力快速增大而外侧接触力逐步减小。结论:1)长距离跑后下肢关节负荷出现幅值及重新分布的变化,表现出增大的髋关节伸及外展力矩,减小的膝关节屈力矩及增大的踝关节跖屈力矩,可能与足姿态变化及相关肌群肌力有关;2)长距离跑后膝关节内侧接触力快速增大且与量化的足姿态参数呈线性关系;3)通过足姿态参数指标可快速预测膝关节接触力的变化,跑者可实地评估足姿态参数以预测膝关节负荷的幅值和分布变化,辅助科学合理的安排跑量及训练计划,降低跑步相关的膝关节损伤风险。

Objective:To reveal the changes of loads in the lower extremity,foot postures and knee contact force and validate their correlations after long-distance running in runners,aiming to provide easily-measured parameters for rapid prediction of loads on knees.Methods:OpenSim subject-specific modelling was used to analyse lower extremity joint angles,moments and knee contact forces after 5-km running.One-dimensional statistical parametric mapping was used to test the statistics.Machine learning algorithm was developed,trained and tested by using foot posture index and total,medial and lateral knee contact forces.Results:1)After 5-km running,the significantly increased angles of hip extension in push-off,hip external rotation in the early stance and ankle plantarflexion during stance were found;2)Significantly increased moments of hip extension during stance,abduction in early stance and ankle plantarflexion during stance were found;however,knee flexion moment was decreased significantly in early stance and push off;3)Total knee contact force was reduced in the push off and lateral knee contact force was reduced in early stance and push off,but medial contact force was significantly increased in early stance;4)The machine learning algorithm showed that the total and medial knee contact forces were increased,but lateral knee contact force was reduced during foot pronation.Conclusion:1)After 5-km running,the lower extremity showed magnitude and redistribution in the loads,which exhibited increased hip extension and abduction moments,reduced knee flexion moment and increased ankle plantarflexion moment,all of these changes may be related to foot posture changes and relevant muscle strength;2)Medial knee contact force was increased significantly at post 5-km running and showed linear correlation with foot posture index scores;3)Rapid prediction of knee contact force could be made by using foot posture index,and runners could evaluate foot postures on-site and estimate the magnitude and distribution of knee loads after running,which could assist with drawing up the running plan and distance,and reduce knee injury risks in running.