Preliminary study for the assessment of physical activity using a triaxial accelerometer with a gyro sensor on the upper limbs of subjects with paraplegia driving a wheelchair on a treadmill.

Objective: This study aimed to examine whether, on the basis of the relationship between sensors attached on the upper limbs and energy expenditure (EE) at the time of wheelchair propulsion, there are differences in the measurement of EE depending on the sensor attachment site and whether addition of the angular velocity information to the acceleration value is advantageous. We also aimed to clarify the variables used to estimate EE as well as the estimated error.

Setting: Laboratory of the National Hospital Organization Murayama Medical Center, Japan.

Methods: Six male subjects with spinal cord injuries participated in the study. Each wore sensors at the wrist and the middle upper arm on both sides while driving a wheelchair on a treadmill at three levels: very, very light; very light; and fairly light. Triaxial acceleration, triaxial angular velocity and EE were measured during driving. We analyzed the correlation between EE and acceleration, angular velocity and synthesized values of acceleration and angular velocity at each location using regression, multiple regression and Bland-Altman analyses.

Results: The determination coefficients between EE and the acceleration, angular velocity and synthesized values of acceleration and angular velocity varied from 0.68 to 0.87 at each location. The mean difference between the measured and estimated EE varied from 0.0028 (s.d., 0.0027) kcal min(-1) kg(-1) on the right upper arm.

Conclusion: These findings suggest that combining the synthesized values of angular velocity and acceleration of the motion sensors on the upper limbs might reflect EE during a wheelchair driving activity on a treadmill.