Optogait

The study presented in this article used the data from the TRIPOD—Treadmill, IMU, Pedobarographic and Photoelectric Dataset [12 (link)] study, available upon request for scientific research purposes. The measurements were obtained from 15 young and healthy participants (8 males, 7 females) who walked on a treadmill for two minutes at three different speeds (Table 1). After two minutes, the treadmill slowed down until it stopped. The dataset also included data from seven IMU sensors (Physilog^® 5 IMUSs from Gait Up, Lausanne, Switzerland). The sensors were located on the left and right instep, left and right heel, left and right shank, and one at the sacrum. The data were acquired at a sample rate of 128 Hz. The IMU data for the three different walking speeds are provided in a CSV file (RF.csv).
Additionally, one file (SyncInfo.csv) contains the timestamp of the initial contact. The ground truth was measured by a pressure-sensitive Zebris (zebris Medical GmbH, Isny, Germany) and photoelectric OptoGait (Microgate, Bolzano, Italy) system. In the multilayer 1D convolutional neural network design, only data from the IMU sensor attached to the right foot, instep position, were used. As ground truth, we used the stride length from the OptoGait system, which is provided in a CSV file (OptoGait.csv).

In order for us to determine walking speed, subjects walked at various speeds without an assistant or support, and the most comfortable speed for each subject was set. While patients were walking at a comfortable speed, the step times of both legs were calculated using a gait analysis system (OptoGait, Microgate S.r.l, Bozen, Italy). Based on the calculated step time, the auditory stimulation sound was produced at an increased rate of 10% for the paretic side and 5% for the non-paretic side, rather than at a comfortable speed. Previous studies have shown that gait velocity and symmetry improved when patients walked to a faster acoustic cue than at a comfortable speed. To reduce the asymmetry of both step times, fast acoustic cues of 10% and 5% were applied differently. The auditory stimulation sound was generated using digital audio editing software (GoldWave v5., GoldWave Inc., St. John’s, NL, Canada). We provided different pitch sounds to distinguish the sounds applied to both legs. Measurements were taken every 2 weeks and acoustic cues were given to subjects according to the changed step time.

Running spatiotemporal parameters were measured with the system previously validated for this purpose, OptoGait (OptoGait, Microgate, Bolzano, Italy) [32 (link),52 (link)]. The default settings for the filter 0_0 (Gait R. in: 0 and Gait R. out: 0 filter) were used. This configuration provides the least bias for time parameters in athletic walking [53 (link)]. Spatiotemporal parameters measured for each step during the 30 s uptake interval at 9 km·h⁻¹, 10 km·h⁻¹, and 11 km·h⁻¹ were the contact time (CT, in seconds; time since the foot touches the ground until the toes separate from the ground), flight time (FT, in seconds; time from the take-off of the forefoot to the initial ground contact of the next contralateral support), vertical oscillation (VO, in centimetres; change in the height of the centre of gravity during the run), step frequency (in steps per minute; number of ground contacts per minute), step length (in metres; distance between two successive contacts with the ground, finger-to-finger) and stride angle (in degrees; the angle formed by the tangent of the parabola traced by the foot to the ground during a stride). The theoretical parabola for determining the stride angle was calculated by the system using the stride length and the maximum height of the foot during a stride [15 (link)].

Ground contact time, stride length, and stride frequency were measured using an optical detection system (OptoGait, Microgate Corporation, Bolzano, Italy). Ground contact time is defined as the time span from the first contact of the foot until the take-off of the foot. Stride length is defined as the distance between the heel of two subsequent footprints of the same foot. Stride frequency can be determined from stride length and the constant velocity of the treadmill during the running economy trials. Ground contact time, stride length, and stride frequency were averaged throughout the same time window of the running economy measurements.

Participants were instructed to put their hands on their waist and jump as high as possible, while maintaining their legs shoulder-width apart, with no restrictions on how low they needed to squat prior to jumping. This test was repeated three times with one minute of rest between jumps. Jump height was calculated by flight time, using a jumping gate (Optogait, Microgate, Bolzano, Italy). The highest value obtained during the three trials was used. If the two best measurements differed by more than 10%, participants were asked to perform a fourth trial.

Gait speed was measured once for each participant by optical-sensitive gait analysis (Optogait; Microgate Co., Bolzano, Italy). Each participant was instructed to walk at their normal speed and the mean gait speed was calculated using a software program (Optogait analysis software, version 1.6.4.0; Microgate Co.) during 5 m of walking at a comfortable speed.