Evaluating squat and countermovement jumps based on a wearable accelerometer: preliminary results

Aim: The aim of this study was to propose an easy and affordable tool for the evaluation of Squat and Countermovement jumps, which could be effectively used in sports practice for the assessment of lower limb performance based on the Bosco test approach. The goal was thus to reliably measure jump flight time based the accelerometer signals and thus compute jump height and work performed. Method: We developed a wearable tri-axial accelerometer based on a MEMS sensor and ZigBee wireless communication with a companion software for PC. The device was worn using a strap-on belt positioning the device roughly at the level of L4. In order to verify our estimates of jump’s flight time, we chose to simultaneously record data from our device and from a Quattro Jump force platform (Kistler) on a set of 15 volunteer subjects each performing one set of five Squat (SJ) and one of five Countermovement jumps (CMJ). The force platform signal was thus used as reference measurement to be compared with those derived from our wearable accelerometer. The acquired data was processed using custom developed Matlab functions. Flight time was computed from the force platform data f(t) as the time interval during which the platform was unloaded and its output was therefore near 0, i.e. f(t)\0.1. Based on the scientific literature we considered three different approaches to accelerometer data processing for estimating flight time: (1) takeoff and landing as the last and first samples, respectively, above a low level threshold indicating freefall (AL), (2) takeoff as the last peak preceding the low level interval and landing as above (APL) and (3) takeoff as the last peak preceding and landing as the first peak following the low level interval (AP). Results: The acquired accelerometer data was generally noisy and required preliminary smoothing. We found that the platform-based estimate of flight time was very precise with a standard deviation of the measurements within subjects ranging 4.1–41.6 ms (mean 12.8 ms) in SJ and 4.6–15.1 ms (mean 9.1 ms) during CMJ. The precision decreased using the accelerometer signals as standard deviations raised to 4.2–83.1 ms (mean 26.4 ms) and 1.6–108.0 ms (mean 31.7 ms) for AL, to 7.1–89.6 ms (mean 54.0 ms) and 6.2–116.4 ms (mean 38.5 ms) for APL, to 5.2–102.2 ms (mean 51.1 ms) and 7.0–113.2 ms (mean 38.8 ms) for AP, in SJ and CMJ, respectively. Conclusion: The accelerometer data was relatively noisy most likely due to the oscillations undergone by the sensor during the abrupt movements typical of vertical jumping. Further effort will therefore be devoted to assuring a tight fit of the worn sensor on the body of the athletes. Although generally underestimating flight times, the most reliable approach was the AL algorithm, though its behavior also relies on signal quality. In conclusion, our results are promising that a reliable evaluation of vertical jumps may be successfully performed based on a lightweight MEMS accelerometer worn at the level of lumbar vertebrae, yet further research is needed to grant better signal quality and reliability of the algorithm.