Associate professor Majmaah University Riyadh, Ar Riyad, Saudi Arabia
Objectives: The study aimed to assess various methods of acceleration analysis in capturing changes in walking sway during dual-task exercises.
Design: 26 healthy adults participated, subjected to dual tasks, which consisted of a motor task (walking) and one of five cognitive tasks. A smartphone-based accelerometer was affixed to each participant's lower back to record center-of-mass acceleration in both anterior-posterior (AP) and mediolateral (ML) directions during the tasks. Various computation methods, including RMS, NPL, mean absolute acceleration, and Peak to Peak acceleration, were applied to the recorded data.
Results:
All computation methods except for mean frequency of acceleration captured significant effects of dual-tasks on sway.
AP and combined (2D) sway directions detected changes in walking sway during the different dual-tasks, while ML direction did not.
The 2Hz low-pass filter was less effective than the 3.5Hz filter in capturing these changes.
Both original acceleration (Acc) and mean referenced acceleration (AccX) captured significant changes in walking sway between different dual-tasks. The necessity of using Acc vs. AccX depended on the computation method.
Conclusions: For detecting changes in walking sway during dual-tasking, it's recommended to compute RMS, NPL, or mean absolute acceleration using the mean referenced AP and 2D acceleration data filtered with a 3.5Hz low-pass filter. Additionally, Peak to Peak acceleration is effective when using non-mean referenced acceleration data in the AP direction.
