A Radar Sensor for Automatic Gait Speed Analysis in Walking Tests

Experimental Verification of Micro-Doppler Radar Measurements of Fall-Risk-Related Gait Differences for Community-Dwelling Elderly Adults

“…We used the process employed in our previous study [33] to extract the upper envelope v u (t) (significant peaks corresponding to the maximum Doppler velocity for each time t), lower envelope v l (t) (significant peaks corresponding to minimum Doppler velocity for each t), and mean envelope of v m (t) (power-weighted mean velocity for each t), as shown in Figure 2. V u (t), v l (t), and v m (t) correspond to leg-forward motion, motion of legs in contact with the floor, and body motion, respectively (information of the arm motions is less, or not, included in these envelopes because the powers of the reflected echoes of the arms were relatively weak and the previous study confirmed that v u (t) almost corresponds to the movements of toes [20,25]). These envelopes were used to extract the following four gait parameters, which were determined to be efficient for the fallers/non-fallers classification in our previous simulation study [32,33]:…”

“…The figure clearly shows the features of the Doppler velocities of the walking motion [24]. For example, the maximum velocity component for each time corresponds to body velocity in gait and includes information on walking speed [25]. We used the process employed in our previous study [33] to extract the upper envelope v u (t) (significant peaks corresponding to the maximum Doppler velocity for each time t), lower envelope v l (t) (significant peaks corresponding to minimum Doppler velocity for each t), and mean envelope of v m (t) (power-weighted mean velocity for each t), as shown in Figure 2.…”

“…Furthermore, as indicated in Table 5, our achieved accuracy and sensitivity values were acceptable compared with the accelerometry-based methods. Next, we compared with other studies on the radar-based gait analysis [24,25]. These conventional studies proposed methods for efficient gait parameter extraction from MDR data.…”

“…The effectiveness of MDR for gait classification has been verified in rehabilitation and hospital applications [21][22][23]. In addition, detailed gait analysis methods based on biomechanical approach using MDR have recently been developed [24,25]. In recent years, machine-learning-based approaches that use MDR data have been employed to recognize various types of motion [26,27].…”

“…In [24,25], general gait parameters were extracted. In contrast, our study extracted the gait parameters related to the fall risks clarified in our study [32,33] and verified the effectiveness of the extracted parameters for the faller classification.…”