A preliminary study on the employment of inertial sensors for wheelchair basketball classification: an investigation into sensor positioning

Introduction
Previous studies have explored the use of inertial sensors to evaluate wheelchair basketball athletes and support their classification processes; however, prior studies have primarily used sensors positioned on wheelchairs, and overlooked trunk and wrist placements, which are key areas for classification. This study investigates the optimal placement of sensors on bodies and wheelchair frames among athletes.

Material and methods
Ten athletes from different basketball classes were recruited. Three inertial sensors were positioned on the trunk, right wrist, and rear region of the wheelchair. The athletes completed the 20-meter speed test and the Illinois agility test. Data on acceleration and speed (inc. data skewness, kurtosis, and RMS) across three axes (x, y, z) were compared with functional classes.

Results
The 20-meter speed test identified six significant correlations in the anteroposterior axis (n = 6), including three accelerometer variables and one gyroscope. In contrast, the Illinois agility test identified twelve (n = 12), including accelerometer data (e.g., linear acceleration, r = -0.86, p < 0.01) and gyroscope data (e.g., angular velocity, r = -0.77, p < 0.01). The strongest correlations were demonstrated by sensors on the wheelchair (n = 8), followed by the wrist (n = 6) and trunk (n = 4). The anteroposterior axis presented the most correlations (n = 12).

Conclusions
Our findings highlight the importance of sensor placement, test type, and movement axis in athlete classification. The wheelchair appears to be the most suitable location for IMU sensors, providing valuable data to distinguish movement patterns between low- and high-class athletes in the classification process.