運用慣性感測器之外在負荷估計換氣閾值跑步速度

Abstract

前言：精準監控換氣閾值速度有助於規劃與執行有效率的跑步訓練。然而，目前在實驗室利用氣體分析監控換氣閾值速度是昂貴、耗時且不方便的。目的：運用慣性感測器之外在負荷估計在跑步機上進行漸增負荷跑步時的換氣閾值速度。方法：六顆慣性感測器分別放置於29 位業餘跑者的下背部、上背部、雙腳脛骨和雙手手腕。受試者在跑步機上進行漸增負荷跑步測試，蒐集並分析換氣閾值速度、最大有氧速度、慣性感測器外在負荷轉折點速度、心率與自覺用力係數。運用斯皮爾曼等級相關分析換氣閾值速度與慣性感測器外在負荷轉折點速度的相關性。應用多元逐步迴歸根據慣性感測器外在負荷轉折點速度建立換氣閾值速度之預測方程式。結果：14個慣性感測器外在負荷第一轉折點速度之中的2至9個與第一換氣閾值速度之間有中度到很高度的顯著相關 (r = .37–.72, p< .05)。此外，在14個慣性感測器外在負荷第二轉折點速度之中的11至14個與第二換氣閾值速度之間有中度到極高度的顯著相關 (r = .38–.90, p < .05)。放置在身體六個不同部位的慣性感測器之估計換氣閾值跑步速度迴歸方程式提供中度到高度的預測準確性 (R: .49–.94, adjusted R2: .22–.87, SEE: 0.43 to 0.91, MAPE = 3.14– 6.35%, p < .05)。迴歸方程式估計第二換氣閾值速度相較估計第一換氣閾值速度有較好的預測表現 (R: .70–.94, adjusted R2: .45–.87, SEE: 0.43 to 0.88, MAPE = 3.14–5.49%, p < .001)。下背部慣性感測器的迴歸方程式 (R = .75–.94, adjusted R2 = .53–.87, SEE = 0.43–0.70 km·h⁻¹, MAPE = 3.14–5.28%, p < .001) 能提供較佳的預測結果 。結論：運用慣性感測器所測得的外在負荷或許可以提供一種即時估計換氣閾值跑步速度的替代方式。研究中所建立的迴歸方程式也可能可以應用於估計跑步有氧能力與開發訓練負荷演算法。

Background: Precisely monitoring ventilatory threshold (VT) speeds is essential for designing and implementing efficient running training programs. However, monitoring VT speeds via gas exchange in lab setting is costly, time-consuming, and inconvenient. Purpose: The aim of this study was to explore the estimation of VT speeds using external loads (ELs) from inertial measurement units (IMUs) during incremental treadmill running. Methods: Twenty-nine recreational runners were recruited and six IMUs were placed on the low back (LB), upper back (UB), bilateral shins, and bilateral wrists of the participants. The participants performed a maximal incremental treadmill running test to collect and analyze VT speeds, maximal aerobic speed, IMU EL-based breakpoint (BP) speeds, HR, and RPE. Spearman's rank correlation analysis was conducted to assess relationships between VT speeds and IMU EL-based BP speeds. Multiple stepwise regression analysis was used to develop estimation formulas for VT speeds derived from IMU EL-based BP speeds. Results: Between two and nine of the 14 IMU EL-based BP1 speeds showed significant, moderate to very high correlations with VT1 speed (SVT1) (r = .37–.72, p < .05). Additionally, between eleven and fourteen of the 14 IMU EL-based BP2 speeds exhibited significant, moderate to extremely high correlations with VT2 speed(SVT2) (r = .38–.90, p< .05). Estimation formulas for VT speeds using IMUs placed on six different body parts have demonstrated moderate to good performance (R: .49–.94, adjusted R2: .22–.87, SEE: 0.43 to 0.91, MAPE = 3.14–6.35%, p < .05). The regression models for SVT2 have shown better predictive performance than those for SVT1 (R: .70–.94, adjusted R2: .45–.87, SEE: 0.43 to 0.88, MAPE = 3.14–5.49%, p < .001). The estimation equation using LB IMU EL-based BP speeds (R = .75–.94, adjusted R2 = .53–.87, SEE = 0.43–0.70 km·h⁻¹, MAPE = 3.14–5.28%, p< .001) has been revealed to be more accurate for recreational runners. Conclusions: IMU EL-based monitoring may provide an alternative approach for the real-time estimation of VT running speeds. The proposed equations may be applied to estimate aerobic running performance and develop training load monitoring algorithms.