發展慣性感測器監控跑步下肢勁度的量測方法

Abstract

前言：下肢勁度對於長距離跑步運動而言是一項重要的生物力學參數。過程中跑者可以透過調整跑姿及步態來改變下肢勁度，若能保持較高的下肢勁度，有利於減少能量消耗。然而目前可穿戴式裝置所量測的跑步力學參數多數無法反映跑步效率。目的：本研究欲透過慣性感測器發展評估跑步下肢勁度的量測方法。方法：招募 20 名業餘跑者，於力板上以固定速度慢跑，並在跑者身上黏貼光點及配戴一顆慣性感測器於骶骨進行資料收集。兩儀器計算結果以皮爾森績差相關檢驗，並以配對t檢定比較差異，後續計算不同速度下的誤差及建立線性迴歸方程式修正偏差。結果：兩儀器間所有參數皆達顯著高度相關，觸地時間 (contact time, CT) r= .908；垂直地面反作用力峰值 (vertical ground reaction force peak, vGRF peak) r= .953；垂直位移 (Δy) r= .814；腿部壓縮量 (ΔL) r= .804；垂直勁度 (vertical stiffness, Kvert) r= .738；腿部勁度 (leg stiffness, Kleg) r= .732。均方根誤差顯示出小的誤差CT = .016 sec；vGRF peak = 88.960 N；Δy = .006 m；ΔL = .014 m，Kvert = 2.988 N/m；Kleg = 1.312 N/m。並且透過布萊特奧特曼圖發現CT (bias= -.042 sec)、Δy (bias= -.001 m) 及ΔL (bias= -.028 m) 有低估的情形，而vGRF peak (bias= 36.408 N)、Kvert (bias= 1.262 N/m) 及Kleg (bias= 2.845 N/m) 則呈現高估。透過迴歸方程式修正後，CT (bias= 0 sec)、vGRF prak (bias= .071 N)、Δy (bias= 0 m)、ΔL (bias= .001 m)、Kvert (bias= -.083 N/m) 及Kleg (bias= -.022 N/m)。結論：本次研究發現IMU量測結果與實驗室儀器有相似之趨勢，或許可以做為室外量測的替代工具，提供更符合真實情境及豐富的資訊給跑者及教練，作為調整跑步策略及訓練安排的依據。Background: Lower extremity stiffness is crucial in long-distance running, impacting biomechanical efficiency. Runners adjust posture and gait to optimize stiffness, reducing energy consumption. However, current wearable devices often inadequately capture these biomechanical parameters, limiting accurate assessment of running efficiency. Objective: This study aims to develop a method for evaluating lower extremity stiffness during running using inertial measurement units (IMU). Method: Twenty amateur runners jogged at fixed speeds on a force plate with reflective markers and an IMU (sacrum) for data collection. Pearson correlation coefficients assessed relationships between both instruments. Paired t-tests compared differences, errors were calculated across speeds, and linear regression equations corrected biases. Result: All parameters between the two instruments showed significant high correlations: contact time (CT) r= .908; vertical ground reaction force peak (vGRF peak) r= .953; vertical displacement (Δy) r= .814; leg compression (ΔL) r= .804; vertical stiffness (Kvert) r= .738; leg stiffness (Kleg) r= .732. Root mean square errors indicated small discrepancies: CT = .016 sec; vGRF peak = 88.960 N; Δy = .006 m; ΔL = .014 m; Kvert = 2.988 N/m; Kleg = 1.312 N/m. Bland-Altman plots revealed underestimations in CT (bias= -.042 sec), Δy (bias= -.001 m), and ΔL (bias= -.028 m), while vGRF peak (bias= 36.408 N), Kvert (bias= 1.262 N/m), and Kleg (bias= 2.845 N/m) showed overestimations. Following regression equation corrections, biases were adjusted for CT (bias= 0 sec), vGRF peak (bias= .071 N), Δy (bias= 0 m), ΔL (bias= .001 m), Kvert (bias= -.083 N/m), and Kleg (bias= -.022 N/m). Conclusion: This study found that lower extremity stiffness indicators measured by IMU closely match laboratory results. IMU could thus serve as alternative tools for outdoor measurements, providing valuable data for adjusting running strategies and training plans.